Talk:Wikivoyage/Migration/FAQ
Next steps
editThis is a brainstorm of next steps for the migration - please add/update anything I've forgotten or misstated:
- Demonstrate a consensus to move the current Wikitravel dump(s) to Wikivoyage. As far as I'm aware there is no dissent to this move, but Wikivoyage would like assurance that the community consents to a move.
- Determine what to migrate - at a minimum English Wikitravel will be migrated, but dumps are available for all current Wikitravel language versions.
- Determine what to do about Wikitravel shared. Wikivoyage already has their own "shared" repository, so discussion needs to take place about how to migrate Wikitravel images to Wikivoyage.
- Determine any legal or other issues related to content import. For example, references to Wikitravel and Wikitravel Extra will likely need to be scrubbed, and this data scrub can take place prior to import (in which case history will be changed) or after import (in which case a bot will need to be created).
- Determine technical challenges. For example: Ryan has a backup of XML and image files, but will need to figure out the specifics of an import with Wikivoyage's technical team. Also, questions remain about how user accounts can be linked (for example, ensuring that import edits attributed to a Wikitravel user are also attributed to the same Wikivoyage user account).
- Determine a schedule. I assume we should expect that the import will need to be done more than once (ie a dry run then the real thing) since with such a large volume of data it's unlikely that the first attempt will be 100% successful.
I think if we can first figure out what the important questions are that it will then be much easier to organize the discussions around answering those questions in the near future. Thoughts? Additions? Corrections? -- Ryan • talk • 01:30, 16 August 2012 (UTC)
1. Consensus
editPlease, voice your opinion here
Four :en WT admins have explicitly stated that they want to move the WT dumps to Wikivoyage, and to do so right away (Peterfitzgerald, Wrh2, Jpatokal, and Shaund). I sent out an email 15 hours ago to all admins, asking them to speak up or forever hold their peace. No one has responded. I think we should balance the desire to have a more explicit consensus, rather than an implicit one, with the urgency of setting this up to preserve the still very large number of good edits being made on WT/en. --Peter Talk 15:09, 16 August 2012 (UTC)
- That's a really important point. I think the worst thing we could do is to split up the stream of mass exodus from wikitravel into those who feel comfortable with WV and those who don't. Either there is an explicit very large majority that supports the migration to WV as an intermediate solution or Ryan has to keep mirroring the active WT wikis until the WMF gives green light for migration under their roof. -- Hansm 17:02, 16 August 2012 (UTC)
- I don't think anything will get split up. Everyone aware of the migration has basically stopped contributing now to Wikitravel, and the quality of that site is degrading, which will make continued mirroring less desirable. If there is any split, it would be between people contributing here and not contributing at all. But those not contributing are already not contributing. The urgency for setting up :en here is that, while Wikitravel goes neglected and un-mirrored, we are losing the good contributions that are made. --Peter Talk 18:39, 16 August 2012 (UTC)
- Erik states that moving to WV will make the final move to the WMF easier. Jmh649 05:16, 17 August 2012 (UTC)
- I have voiced my own support over at Interest in starting a new language version but am happy to qualify it here as well. A stand out concern is that of losing good quality new emerging content due to the rising tide of dross entering WT at the same time. Good content is likely to be increasingly smothered under a layer of encroaching dross until it eventually becomes hopelessly obscured. The current predicament also presents a considerable risk of losing the goodwill and interest of ongoing contributors who are unaware of the of the somewhat arcane behind the scenes machinations of all this. We could consider this as a rescue mission as WT is likely to become a zombie unless IB overcome what appear to be insurmountable systemic inadequacies. Moving the project here temporarily may look like a fork but we should perhaps view it as prudent triage, we would be putting the project on life support whilst the remedy is clarified.--felix 13:47, 18 August 2012 (UTC)
- Erik states that moving to WV will make the final move to the WMF easier. Jmh649 05:16, 17 August 2012 (UTC)
- I don't think anything will get split up. Everyone aware of the migration has basically stopped contributing now to Wikitravel, and the quality of that site is degrading, which will make continued mirroring less desirable. If there is any split, it would be between people contributing here and not contributing at all. But those not contributing are already not contributing. The urgency for setting up :en here is that, while Wikitravel goes neglected and un-mirrored, we are losing the good contributions that are made. --Peter Talk 18:39, 16 August 2012 (UTC)
- I support migrating the database dumps here. I am Wikitravel user ravikiran_r Ravikiranr 17:33, 22 August 2012 (UTC)
2. Language versions
editPlease correct my guesses for the new project namespaces on tech:New namespaces for language versions -- Hansm 17:07, 23 August 2012 (UTC)
I would ask that you either migrate all language versions now, or continue to run backups for language versions other than English. From our side, we can maintain :ru on WT as long as necessary (and it is not difficult anyway), but I personally would like to say farewell to WT right now -) Atsirlin 06:05, 16 August 2012 (UTC)
- Have no idea, how active these language versions are. I think we can leave out some. They can be started from a scratch later. Just keep the articles of Wikitravel namespace. So we can use them for a possible restart at the WMF. How about the longer established and still a bit active ones? Can we import them directly to the WMF then? I think we have limited human ressources and should focus on en: first. What languages are worth to be transfered to the WMF? Have you talked about that? -- DerFussi 06:15, 16 August 2012 (UTC)
- Thanks for the fast information about ru: -- DerFussi 06:16, 16 August 2012 (UTC)
- Yes, sure. It is up to people with tech experience to decide. Could they say about their preference:
- i) import one language version at a time?
- or ii) import all viable language versions, in order to keep data integrity, inter-language links, etc.?
- Then we know whether or not the language issue should be discussed further. Atsirlin 07:40, 16 August 2012 (UTC)
- As far as I understand the situation on WT, continuing making backups is the crucial point. Since IB has blocked the API, there is no more way for any automatic backup from outside of IB's servers. Thus, it would not make too much sense continuing maintaining :ru or any other language version on WT as a lot of manual work would be needed to transfere this contributions, later.
- Setting up a running wiki from XML-dumps is always some kind of mess, be it too long dump files, some very rare character encoding problems or whatever. From the technical point of view, I would prefere to migrate all active language versions at once. This way, we could apply our throw-away-repair-scripts to all language versions at one time and save the time for archiving them for later use and understanding them again.
- -- Hansm 07:50, 16 August 2012 (UTC)
- Thanks for the comment. Ryan mentioned that the API is enabled again, and new backups are possible. But if you prefer to move everything at the same time, let's discuss which language versions are still active and worth moving. Atsirlin 08:45, 16 August 2012 (UTC)
Let's try to summarize the present status of different language versions: number of active admins/regular contributors, total number of articles, number of meaningful articles, willingness to move
ar 165 articles. No meaningful edits in the last 100 days
ca 614 articles. 1 meaningful edit in the last 100 days
de skipped
eo 502 articles. No meaningful edits in the last 100 days
es 1928 articles
fi 1645 articles
fr 2518 articles
he 96 articles. He has been almost entirely spam for some time now. --Peter Talk 14:43, 16 August 2012 (UTC)
hi 235 articles. No meaningful edits in the last 100 days
hu 672 articles
it skipped
ja 5116 articles
ko 266 articles. No meaningful edits in the last 100 days
nl 3043 articles 3 active admins
pl 2527 articles
pt 2900 articles
ro 570 articles
ru 2 active admins and 4-5 regular contributors. 1594 articles. My rough estimate is that we have at least 100 meaningful articles, with many of them covering distant places in Russia. It is a piece of rare and rather specialized travel information. We have also drawn a bunch of original maps. All active contributors are willing to move. Atsirlin 13:18, 16 August 2012 (UTC)
sv 1422 articles
zh 513 articles
- I think Atsirlin's list is almost exactly right regarding which language versions are worth backing up here—I think :he is also not necessary. We can keep the back ups as of now for ar, ca, eo, hi, and ko, in case new contributors want to get working on them for real post-migration. I do think all others, except :it and :de should be moved here. If the API is working, I think it would make sense to run a last update of those language versions before migrating them here.
- One of the best reasons to migrate the other language versions here might just be to save them from the spambot rampage. :Fr is a version I have worked on quite a bit that has a ton of good content, but will require a serious effort to clean up the damage of the past several years, in preparation for the migration. That would be much, much easier to do here. --Peter Talk 14:43, 16 August 2012 (UTC)
- Two comments:
- I'd like to see someone from each language version to be migrated (and ideally multiple people) request migration before anything is moved. We need to have at least one person active to have a hope of the site being successful.
- Similarly, I can re-enable the spider on a language-by-language basis if someone from that language version requests it. I'll re-enable Russian for now, and if other language versions remain active on Wikitravel then those users can also request a backup; however, I don't want to continue spidering language versions that aren't being patrolled, since that just means more spam to deal with later.
- -- Ryan • talk • 15:11, 16 August 2012 (UTC)
- Two comments:
Language versions to import
editBefore a language version will be imported to Wikivoyage we would like to ensure that there will be at least one active administrator who can patrol it and help new users. Please comment at Interest in starting a new language version if you are willing to fulfill this role.
- Previous table removed - please comment at Interest in starting a new language version
Hans, do you want to launch a poll for other language versions as well? Although possible, this may not be viable, because we will hardly get more than 4-5 votes per language. Moreover, many of the contributors are not fluent in English. Should we assume that the table here (i.e., 1 person/admin per language) is a sufficient basis for the migration? Atsirlin 12:37, 18 August 2012 (UTC)
- I think the (current) result of the en: poll is so clear that we can assume that there is no real opposition even against other language versions. The question is for which language versions it would make sense to activate them. If there is only one or two admins or contributors, this seems to be to less. I would like to require a minimum of 5 contributors in all and 3 of them should be willed to become an admin, no matter if already admin on WT or not. Interested contributors can sign on Interest in starting a new language version. -- Hansm 17:35, 19 August 2012 (UTC)
- OK. I agree that we should appreciate the existing rules of WV-) But could you please post the same information somewhere on Interest in starting a new language version instead of Ryan's suggestion to vote here? Presently, it is quite messy because there are at least two places to vote. Atsirlin 17:56, 19 August 2012 (UTC)
- Unfortunately, you are right. It's really somewhat messy ;-( But let me have some supper, first. -- Hansm 18:00, 19 August 2012 (UTC)
- How set in stone are the number of contributor requirements for the import? For :fr, I know that even if it was just Joelf and I working on it, we could really clean things up in preparation for a move. Otherwise we'll have a pretty bad looking site for moving to the WMF. I suppose I probably could drum up enough contributors for most of the language versions above, but I already have my hands pretty full at present, trying to contact and then coordinate with 600+ users. --Peter Talk 18:46, 19 August 2012 (UTC)
- Maybe we could try putting a site notice on the other languages? Sumone10154 02:27, 20 August 2012 (UTC)
- We certainly can, although IBobi has tried to prevent this before (when Jpatokal added one to :fi). I assume he'll be reading all of this too. --Peter Talk 04:38, 20 August 2012 (UTC)
Atsirlin, I have asked the Wikivoyage board members to explain the policy and procedure for non-en: language versions. For now, I will concentrate more on the technical issues. -- Hansm
Btw, I have announced the running poll for the en: migration right there where I expect it to be read by en-Wikitravellers: In the pub on WT/en. It's still there, althought posted on Saturday morning ;-) I think this would be a good way for other languge versions, too. But before, Wikivoyage should be prepared with clear signposts telling new contributors where to go, what to read and where to sign. Unfortunately, this is not yet the case for non-en: language versions.
-- Hansm 09:35, 20 August 2012 (UTC)
- Yes, of course, we do our best to disseminate the information. But sometimes it does not work. For example, I am unable to change MediaWiki:Sitenotice at WT/ru. I presume that it was silently blocked to prevent any edits. --Atsirlin 17:56, 20 August 2012 (UTC)
- I've added a notice on the pub at WT/fr, but I'm also unable to edit the sitenotice there. Sumone10154 22:51, 20 August 2012 (UTC)
Could we remove the voting table on this page and move the votes to Interest in starting a new language version? There are currently two places to vote. Sumone10154 22:56, 20 August 2012 (UTC)
- I've removed the voting table. -- Ryan • talk • 23:36, 20 August 2012 (UTC)
Please correct my guesses for the new project namespaces on tech:New namespaces for language versions -- Hansm 17:07, 23 August 2012 (UTC)
3. Shared:
edit- Still feel a bit uncomfortable with the missing license tags. Ok. At your upload page is an absolutely clear statement: If somebody uploads a file, at least cc-by-sa-3.0. Is there a license tag automatically added to the description page? How about the ones before your license upgrade to cc-by-sa-3.0? Is there a point of time? How to handle them?
- There are files in your language versions as well [1]. What should we do with these images? How to handle the unlicensed images at the language versions? ( e.g. [2])
- Two ideas 1. we should use separated repositories first and seize this chance to check your files. Maybe your files that are licensed and categorised properly can be moved to our shared: automatically. 2. Importing everything to our shared. And all files with something missing (license, category, source (some may be importet from commons, flickr ...)) will get a needs a check tag. ....
- Have you files from Commons, Flickr ... uploaded to your wikis? Are they checked (source, license ...)
- We should use only one repository for uploading then. Its usefull to have only one shared: to be transfered to the WMF. And this is a chance for a revision. -- DerFussi 06:15, 16 August 2012 (UTC)
- I'm pretty sure that we did not upgrade image licensing, as we direct re-users to look at image pages for image-specific licensing (since, obviously, not all images are CC-by-SA 3.0). I could be being senile though. It may be possible to automatically add a CC-by-SA 1.0 tag to images uploaded before we had licensing tags (when we simply had a message instructing uploaders that their work will automatically be so licensed if they choose to upload it to Wikitravel), but again, I don't know how feasible that is.
- These should be deleted or moved. Ryan had been doing an enormous amount of work on this for :en. I haven't checked to see how big a problem this will be for other language versions, aside from :ru, where I know it is not a problem. --Peter Talk 16:27, 16 August 2012 (UTC)
- Well, :ru has 470 images (I was also oblivious of that). Should we sort them out before the migration, or this can be done later? Atsirlin 17:20, 18 August 2012 (UTC)
- That makes sense.
- Yes, and sort of. We do have a Template for flickr review [3], but we have not had the manpower to use it extensively, aside from a sustained, but Quixotic effort some years ago. Commons files can always be verified by clicking the link. But our basic ideal has been to trust users unless they appear untrustworthy. That's how contributions are monitored in terms of self-made images, as well as original text contributions to Wikitravel. Copyright violations are usually easy to spot. Sometimes they are not, but we can only do the best we can.
- That also makes sense. --Peter Talk 16:27, 16 August 2012 (UTC)
- Is it possible to move directly to Wikimedia Commons and just make that the defacto shared? Or do we need to move to WV to check the images for copyright issues first? Jmh649 05:23, 17 August 2012 (UTC)
- WT Shared is pretty huge, and is going to take a while to sort out, especially as we'll want to keep a good number of the files local, rather than on Commons. --Peter Talk 05:25, 17 August 2012 (UTC)
- But the ones that can be moved to commons should be yes? Jmh649 05:48, 17 August 2012 (UTC)
- WT Shared is pretty huge, and is going to take a while to sort out, especially as we'll want to keep a good number of the files local, rather than on Commons. --Peter Talk 05:25, 17 August 2012 (UTC)
- I think the long-term plan is to move almost everything to Commons, although as Peter noted there is a significant amount of work to do in order to make that happen (vetting licenses, ensuring proper attribution, etc), so importing into Wikivoyage and then moving images on a case-by-case basis might make the most sense. -- Ryan • talk • 22:17, 17 August 2012 (UTC)
- For reference, here's the count of images for each language version:
Language Number of images ca 34 de 1387 en 7839 eo 8 es 798 fi 114 fr 1938 he 82 hi 1 hu 123 it 2446 ja 1724 ko 1 nl 294 pl 1012 pt 547 ro 35 ru 469 shared 33979 sv 459 zh 63
- Given the large amount of files to deal with I'd suggest importing them as they are to Wikivoyage and then figuring out a process for moving as many as possible to Commons. -- Ryan • talk • 17:46, 18 August 2012 (UTC)
Will we keep the Wikivoyage image shared repository and WT Shared separate when we move it here? It seems like it might be a mess otherwise—at least until we've spent some real hard work sorting things out. --Peter Talk 17:55, 18 August 2012 (UTC)
- I think that would be a clean way of allowing the Wikitravel shared repository to be cleaned up and slowly moved to commons, but hopefully one of the Wikivoyage admins could comment as to whether it is technically possible or not. -- Ryan • talk • 18:39, 18 August 2012 (UTC)
3.1 Redirects
editJust about to check and tidy up Cambodia on WT/shared. WT seems to have redirects from main namespace to the categories (e.g. Cambodia redirects to Category:Cambodia. All these redirects misses a colon inside the link in front the word "category". Thats why this kind of a loop is built and you see these italic written categories. Normally you dont need the redirects at all. On shared the categories should be found automatically. So maybe the main namespace should be left out during the import. -- DerFussi 08:49, 17 August 2012 (UTC)
- Those redirects should not be there—those are just mistakes. You are right, though, that we should probably run another back up of shared, since it has not degraded like other versions. --Peter Talk 15:47, 17 August 2012 (UTC)
- If someone is willing to go through and delete the spam pages on shared then I'm happy to re-run the spider on it. Please contact me once a cleanup is done. -- Ryan • talk • 15:52, 17 August 2012 (UTC)
- Ok. So there are just some mistakes. I've just started to tidy up the mess on shared: in Cambodia. Ok. Then I will do some more changes on WT shared: after the diner tonight. I'll give you a sign then. -- DerFussi 15:56, 17 August 2012 (UTC)
- Work to be done and calls to be done... At least I've been able to tidy up Cambodia and check all other edits. Just checked Myanmar ... lot of work to do .. depends on the start of import. I'll be able to clean up the country of Myanmar on Sunday. If we start earlier.... you can start the spider. -- DerFussi 21:21, 17 August 2012 (UTC)
- The spider is running now against shared, and jamguides.com should be up-to-date with the new content tomorrow. -- Ryan • talk • 22:17, 17 August 2012 (UTC)
Not everything is updated. After the migration a will work on these countries again to make more compatible to our shared: and commons: concerning spelling and hierarchy. -- DerFussi 15:12, 21 August 2012 (UTC)
- Sorry, I should have updated my comment - I forgot to disable the cron job for reloading jamguides.com, so while the spider backed up your recent changes to shared, the reload of jamguides.com started before the spider had completed. I'll reload jamguides now, although it takes about ten hours to complete. -- Ryan • talk • 15:14, 21 August 2012 (UTC)
- The update and reload finished last night - jamguides.com now has the latest data from the last backup we were able to grab before IB disabled the Mediawiki API. -- Ryan • talk • 21:52, 22 August 2012 (UTC)
Datestamp
edit- What is the datestamp on that? --Peter Talk 22:15, 22 August 2012 (UTC)
- All times are Pacific. The API was disabled by IB on 09-August but re-enabled (briefly) a week later. A full backup was requested for ru: and shared:, and one-off backups were requested for about 30 files on en:
ar Aug 8 13:00 ca Aug 8 14:00 de Aug 8 15:00 en Aug 17 22:49 (partial backup - last full backup 02-August) eo Aug 8 16:00 es Aug 8 17:00 fi Aug 8 18:00 fr Aug 8 19:00 he Aug 8 20:00 hi Aug 8 21:00 hu Aug 8 22:00 it Aug 8 23:00 ja Aug 9 00:00 ko Aug 9 01:00 nl Aug 9 02:00 pl Aug 9 03:00 pt Aug 9 04:00 ro Aug 9 05:00 ru Aug 16 08:11 (full backup requested by Atsirlin) shared Aug 17 15:14 (full backup requested by DerFussi) sv Aug 9 07:00 zh Aug 9 08:00
- -- Ryan • talk • 22:28, 22 August 2012 (UTC)
3.2 Import plan
editThe following is a summary of an email exchange I had with Hans:
Me:
- One other thing that I don't think I've seen discussed: what is the plan for the shared repository? Will there be a separate shared repository for the imported sites, or is the plan to merge it with your existing shared repository? My thought is that a separate repository might be easier to manage, and then once we've moved most of the images to Commons we can try to merge the existing Wikivoyage shared with the new (imported) shared. Did you have another plan?
Hans:
- Concerning the file repositoriums, I would perfectly agree with you. Let's make a 1:1 copy of what users are used to work with on WT. i.e. a shared wiki and the language versions with their existing files. Just one thing: I think we should disable the file upload on the language versions and force users to upload their new files either on shared or directly on WM/commons. Also, I would like to keep our existing WV/shared separate from the imported WT/shared. Let's say we call them shared and shared2. It would need a lot of work to merge our WV/shared with the WT/shared. In the near future, we should try to integrate all existing files into WM/commons. But this is a challenge for the community, not for us technicians.
So the plan (barring any changes) is to import the shared: repository from Wikitravel as-is, and keep it separate from the current Wikivoyage shared: repository. As content is migrated into Wikimedia Commons and the shared: repositories here shrink we can revisit merging the two repositories back together. While it might be nice to start out with a merged repository, the technical challenges of doing so appear to be beyond what is possible given our short timeframe for getting the language versions up and running. -- Ryan • talk • 03:23, 23 August 2012 (UTC)
- commons:User:Sven Manguard started a related discussion on the Wikimedia Commons site: commons:Commons talk:Wikitravel Shared transfer task force. -- Ryan • talk • 00:38, 24 August 2012 (UTC)
4. Legal issues
editAttribution
editFor this discussion, we have to refer to the full legal code of CC-by-sa 3.0. §4,c seems to be the most important for us. It is about the "by".
There are 4 points to take care about, labled as (i) to (iv):
- (i): Name of the Original Author
- (ii): Title of the Work
- (iii): URI associated with the Work
- (iv): Credit identifying the use of the Work in the Adaptation
In our case, I assume (i) allows to refer to the so called "Attribution Party", i.e. "Wikitravel", instead of listing each author's name.
I fear we cannot scrub the references to Wikitravel and Wikitravel Extra as this would violate point (iii).
Since we can import the whole version history for each article, the identification of the adaption as required in (iv) would be abvious without any further measures.
-- Hansm 08:36, 16 August 2012 (UTC)
- (i): We have always used the name of the original author, and that information is preserved in both the article history and credits.
- (iii): Maybe it would be possible to automatically add information to article histories to say below which point all edits were made to wikitravel.org? Otherwise, a pointer to our Copyleft could be placed at the top of all article histories to show the same point (since they will all be the date of the final back up). --Peter Talk 14:54, 16 August 2012 (UTC)
- We already have discussed the aspects of attribution before launching de:. The problem is that there is not only the revision history, but also features like the api, the page export and probably some others that all need to provide the correct attribution information. Not to forget the footer of each article that is the only way for giving attribution when an article is printed.
- It would be terribly messy and error-prone to insert this extra info info about edits made on WT on all these features.
- For me, the clearest way still remains to label WT contributors with a preceeding "WT/en" or whatever. This way, not additional changes are needed.
- -- Hansm 16:54, 16 August 2012 (UTC)
- Hi Hans:
- Can you clarify what you mean by "features like the api, the page export and probably some others that all need to provide the correct attribution"?
- As to the footer attribution, Wikipedia currently does not provide footer attribution, even when printing, so are you sure that's legally required? Unless we are distributing printed guides (which we're not) then I don't think we have to worry about printing.
- My assumption was that we shouldn't change user names in history since that's actually a part of the attribution - if someone uses ID "WikiLogin" across multiple projects, that is effectively the only way to identify them and thus the credit required by the CC-SA (see my next point for further thoughts). Since the article history shows all contributors, and since user pages would be imported to provide any additional identification that users have provided, my assumption was that a full history import would thus provide all required attribution.
- To your point above, the section 4c of the CC-SA requires "the name of the Original Author (or pseudonym, if applicable)", so it sounds like login (pseudonym) is considered sufficient.
- -- Ryan • talk • 17:26, 16 August 2012 (UTC)
- Hi Hans:
- 1. There are several features in the api that return the license info and a list of contributors. If someone only looks at the uptput of the queries, he will get missinformed. Sure, as far as I know, there is no way to return a note about the source work, but at least we coult give a hint by labeling the contributors as non-WV users. Also, there is some almost forgotten, api-loke feature for polling license and contributor information, but I do not rember how it is called. Evan has been one of its devellopers.
- 2. Sure, simply because the wikipedia does not need to give attribution to any source work. For them, it's enough to say something like "This article is from the Wikipedia."
- 4. Of cousre, the login name is sufficient. But I assume it would even be sufficient to say "This is from Wikivoyage, based on the source work xy on Wikitravel."
- -- Hansm 17:51, 16 August 2012 (UTC)
- Two comments:
- First, my understanding is that it's not required to credit Wikitravel, but it IS required that we credit the author. "Wikitravel" did not write anything and does not hold a copyright or license on the work, but the user who made the contribution must be credited. If (for example) someone uploads a picture of mine from Wikitravel to Commons, they must credit me, not Wikitravel.
- Second, I'm a bit hesitant to change user ID during import, since that seems like it would have legal concerns due to the CC-SA attribution requirements, and also because a user's contributions would no longer be attributed to them on the new wiki - for example, my Wikitravel contributions would be attributed to wt_Wrh2, or something like that. Using the commons analogy again, if someone copies a photo of mine from Wikitravel to Commons they have to credit "Wrh2", not "wt_Wrh2".
- From a technical standpoint, per [4] it sounds like if an edit was attributed to me (Wrh2) in the XML import, and I then register an account on English Wikivoyage as "Wrh2" that I would then "claim" those edits. If there is a concern about having users "claim" someone else's username, maybe we could pre-register all existing Wikitravel accounts and set up a process where people could prove they are the Wikitravel user? I think I could write a script that would create accounts for all credited Wikitravel users if that would be useful. What are your thoughts? Is that necessary? Are there other options we could pursue besides changing user names for edit histories? -- Ryan • talk • 03:03, 17 August 2012 (UTC)
- From my understanding either the history needs to be imported or a link to the authors in the history needs to occur. It is fairly minimal attribution. Jmh649 05:28, 17 August 2012 (UTC)
- History will most definitely be imported - the XML backups from Wikitravel include complete topic histories for all articles, including user pages. -- Ryan • talk • 06:06, 17 August 2012 (UTC)
- From my understanding either the history needs to be imported or a link to the authors in the history needs to occur. It is fairly minimal attribution. Jmh649 05:28, 17 August 2012 (UTC)
- Two comments:
From the site owner's point of view, i.e. the Wikivoyage Associoation, things are somewhat different then from the users' point of view. The WVA could very easily run into a terrible judical trouble if the CC-by-sa license is not complied to 100%. As a member of the WVA, I only can advise to make judical things rock solid. This means labeling contributors from foreign sites as such, e.g. prefixing their names with something like "WT/en". Also, say the name of the source work and a URL to it on the pages' footers. All other proposals given above sound too delicate to me.
Additionally, we can reserve all WT user names and establish a verification mechanism. At least as long as IB does not block the API again.
-- Hansm 17:54, 19 August 2012 (UTC)
- Hans, I think changing a contribution history by adding "WT/en" to a user name would actually put the WVA in questionable legal territory since the CC-SA requires contributions to be credited to the original author or pseudonym used by the author (from the CC-SA text: "the name of the Original Author (or pseudonym, if applicable)"). I'm not a lawyer, but my reading of the CC-SA is that users must be credited using the name that they originally contributed with. As to crediting Wikitravel, I don't think that is required, but to be 100% safe could we do something like I did on jamguides.com, where the footer of each page says:
- Content on these pages was originally imported from Wikitravel and the original author attribution can be seen by viewing the history tab of each article. All article text is available under the Creative Commons Attribution-ShareAlike 3.0 license, while images are available under various licenses; see each image for details.
- Would that address your concerns? To be clear, I'm raising the issue of changing contributor name because I want to ensure compliance with the CC-SA, and I think changing it would actually put the WVA in danger of not complying with the CC-SA license. -- Ryan • talk • 18:40, 19 August 2012 (UTC)
- We could also affix a date to that message, to show at what point the website changed.
- Also, Wikitravel has always had a message that, if you enter something into the "real name" field when registering, your work will be credited to that name. E.g., my username is easier to type as peterfitzgerald, but I am credited as Peter Fitzgerald [5]. --Peter Talk 18:49, 19 August 2012 (UTC)
- Unfortunately "real name" is not included in the XML export data:
<contributor> <username>Peterfitzgerald</username> <id>12794</id> </contributor>
- While my understanding is that Mediawiki is smart enough to associate the imported "Peterfitzgerald" contribution with your "Peterfitzgerald" account on the new site, I don't know if it will also then credit your full name based on your account settings. If we change the import account to something like "wt_Peterfitzgerald" then that association would be lost, which I think creates a legal concern about properly crediting accounts. -- Ryan • talk • 19:00, 19 August 2012 (UTC)
- (Response to Hans) One further point, the "URI" requirement in the CC-SA is the following:
- (iii) to the extent reasonably practicable, the URI, if any, that Licensor specifies to be associated with the Work, unless such URI does not refer to the copyright notice or licensing information for the Work
- My understanding is that this requirement is only applicable when the contributor specifies a URI with licensing info, such as when someone specifies an image as coming from Flickr or another source. For text contributions, the "URI" would most likely be the user page, since the "licensor" is the contributor and the user page is where they would specify any special licensing for their contributions, and the user page URI is automatically included in topic history. Linking to Internet Brands / wikitravel.org should not be a requirement since they are not the copyright holder or licensor. -- Ryan • talk • 19:22, 19 August 2012 (UTC)
- Concerning the suggestet prefix WT/en. Why we do not use “Wikitravel/en user ” as a prefix in the history? We can use spaces. Can't we? So the original name of the contributor is stated. -- DerFussi 04:56, 20 August 2012 (UTC)
- Hopefully others can weigh in, but it seems to me that any change to the contributor name in the history would violate the CC-SA. Hans' original comment was "In our case, I assume (i) allows to refer to the so called "Attribution Party", i.e. "Wikitravel", instead of listing each author's name" which I think is wrong - Wikitravel is not the author and copyright holder, the contributor is the author and copyright holder. My contributions don't belong to Wikitravel, they belong to "Wrh2", and the CC-SA specifies that they must be credited to that pseudonym (per section 4.c.i of the CC-SA [6]). If "Wrh2" is pre-pended with something else during import then contributions made on Wikivoyage will no longer show up as "belonging" to the original author, and would not link to my user page where any additional licensing info I might have specified is stated (per section 4.c.iii of the CC-SA [7], as noted above). Similarly, Special:Contributions/Wrh2 would show none of my original edits, so they would no longer be credited to me. As a result, I think any effort to pre-pend something to the pseudonym is problematic from both a legal and ethical standpoint. Am I misunderstanding something? -- Ryan • talk • 06:26, 20 August 2012 (UTC)
- See also [8], which has an overview from Creative Commons about how to properly attribute someone. It looks (to me) like the relevant portion is "Cite the author's name, screen name, user identification, etc. It is nice to link that name to the person's profile page, if such a page exists.". That requirement would be covered if we import article history and also import user pages (both things we plan to do) and do not change any of the ids/logins in the article histories. -- Ryan • talk • 07:17, 20 August 2012 (UTC)
I happened to see this discussion and would like to add some points. Peter is not only credited as "Peterfitzgerald"; he is also credited as "12794". However, if you import the edits before creating a "Peterfitzgerald" user on the new project, then the edits will be credited to "0" instead of "12794". If the account "Peterfitzgerald" already exists on the new project when you import the edits, then the system credits the edits to that account under whatever user ID it has on the new project.
For example, take this edit to Greek Wikipedia. Made by me on English Wikipedia and then imported to Greek Wikipedia before I created my account there. Currently credited to user ID 0 since there was no Greek account with that name when the edit was imported. To check the user ID, export the page as an XML file, and search for my username: user ID "0" will be indicated close to the user name.
Crediting a user to "0" has some odd features:
- Article history and diffs have a link to Special:Contributions and to the talk page, but not to the user page (same as for IP edits). Is this a violation of the requirement to provide a URL?
- Using recent versions of Mediawiki, the edits don't appear in Special:Contributions (check here). This may be undesirable. However, I think that things worked differently in the past, so maybe there is a setting somewhere.
- If the account is renamed, any edits credited to user ID "0" will remain under the old user name. In fact, someone else "stole" my user name on Greek Wikipedia at some point, and then I had to usurp the account by requesting renaming. The edit didn't change its attribution when I requested renaming of that account.
Stefan2 03:04, 22 August 2012 (UTC)
- Thanks Stefan. In our test instance I just verified that for the China article Special:Contributions/Wrh2 shows my contributions (my account was created prior to importing the article), but (for example) Special:Contributions/Pashley does not show any results, despite the user being credited via the article history. I don't think that causes a legal issue since the history still shows the user name and would link you to the user page, but it would still be nice if there was a way to get that wired correctly. We'll need to figure out whether that is fixable post-import via some batch process that could run occasionally (hopefully!), whether pre-creating user accounts makes, or whether we can live with this as-is. I don't have time to research further right now, but if anyone has any further input or could link to any Mediawiki documentation that addresses this issue it would be helpful. -- Ryan • talk • 03:25, 22 August 2012 (UTC)
(iii) has as caveat "unless such URI does not refer to the copyright notice or licensing information for the Work", which, strictly speaking here, we might consider it does not. This might be a good question to ask of the WV or WMF counsel (or both). --Kim Bruning 09:58, 23 August 2012 (UTC)
Scrubbing Data
editGeneral Scrubbing Discussion
editAny references to "Wikitravel" will need to be scrubbed from the imported content when we do a migration to Wikivoyage. On jamguides I run a bot [9] AFTER doing the import to perform this scrub so that the contribution history is unchanged. That has the advantage of not changing someone's past contributions, and also showing where content came from - from a legal standpoint that seems safer to me, but others may disagree. So question for those involved:
- Should we scrub the content before or after import? Before means history is changed, after means the history includes references to Wikitravel.
- What should be scrubbed? A few people have mentioned things like changing "Get out" references, but I'd like to tackle that as a separate project from the initial import. I'd propose writing a bot to:
- Change references from "Wikitravel" to "Wikivoyage". I suspect the initial scrub will be overzealous (for example, the current Pub discussion about forking Wikitravel won't make sense after an automated change), but things can be reverted on a case-by-case basis afterwards.
- References to Wikitravel Extra should be removed since there is no Wikivoyage Extra.
- What to do about Wikitravel Press references? Leave them?
- Update any hard-coded wikitravel.org URLs to (I think) en.wikivoyage.org, fr.wikivoyage.org, etc.
- Update any interwiki links as required to match the Wikivoyage format.
A related issue is whether edit summaries need to be scrubbed - I'll investigate further, but if they do then that would obviously need to be done prior to import, unless we are going to modify content in the database later on.
Comments or suggestions welcome. Since the current schedule calls for starting on the technical issues by Wednesday it would be good to resolve the above ASAP. -- Ryan • talk • 19:13, 19 August 2012 (UTC)
- We are winding down with WTP—it's OK to take out the references, along with other instances of Wikitravel. --Peter Talk 19:48, 19 August 2012 (UTC)
- Just a caveat: some languages require language-specific replacements. In :ru, we use "Wikitravel" in page names but "Википутешествие" (+all derivative forms according to different genders) as the name of the project. I should think on how to deal with that. I hope it is irrelevant to other European languages, whereas :ja and :zh are probably not going to be moved. Atsirlin 19:59, 19 August 2012 (UTC)
- Oh, also, can we scrub all references to ==Get out==? Per the approval vote and subsequent discussion here [10], we are ready to change that benighted header to Go next! --Peter Talk 20:38, 19 August 2012 (UTC)
- Yes, sure, but this is again irrelevant to :ru, because we have "Окрестности", which is neither "Get out" nor "Go next", and which I personally would like to keep. Atsirlin 20:48, 19 August 2012 (UTC)
- I'd prefer to minimize the scope of work required for the migration, so it would be better to tackle the "Get out" change after doing the initial import if that's OK. Once a bot is written to change data then updating it to change almost any pattern should be relatively trivial. -- Ryan • talk • 21:21, 19 August 2012 (UTC)
- Now I can be specific. Aside from the standard Wikitravel --> Wikivoyage replacement, :ru will need the following:
- "Википутешестви*" --> "Wikivoyage" (here * means one letter)
- "википутешественник" --> "участник"
- --Atsirlin 17:16, 20 August 2012 (UTC)
- Thanks! If anyone else has similar patterns that need updating please collect them here so that we have them all in one place. -- Ryan • talk • 17:40, 20 August 2012 (UTC)
I'd almost rather leave WTP and Wikitravel Extra references and edit them nicely by hand. Wholesale excision may end up looking really weird. LtPowers 02:23, 21 August 2012 (UTC)
- Murphy's law of "letting the bot do it" says that this always ends up REALLY weird. As far as I'm concerned, the new version is the main/canonical fork of wikitravel. --Kim Bruning 10:00, 23 August 2012 (UTC)
Pre-Import Scrubbing
editTwo requests for those who can help, one technical and one not technical:
- I mistakenly thought that Mediawiki converted namespaces automatically on import (for example: [[Wikitravel:About]] needs to become [[Wikivoyage:About]]), but that's not the case. My
sed
skills are horrid, so can anyone help out with the appropriatesed
expressions for converting namespaces on import? We need to convert the following:- [[Wikitravel:, [[:Wikitravel:, {{Wikitravel:, {{:Wikitravel:
- [[Wikitravel talk:, [[:Wikitravel talk:, {{Wikitravel talk:, {{:Wikitravel talk:
- [[Wikitravel Shared:, [[:Wikitravel Shared:, {{Wikitravel Shared:, {{:Wikitravel Shared:
- [[Wikitravel Shared talk:, [[:Wikitravel Shared talk:, {{Wikitravel Shared talk:, {{:Wikitravel Shared talk:
- ...plus any custom versions for each language version
- If various language versions have their own translated namespace for the "Wikitravel" namespace could someone also help out by listing those with the corresponding namespace to use for Wikivoyage?
Thanks! -- Ryan • talk • 05:26, 23 August 2012 (UTC)
- Regarding #2, the only language versions to have done that are :he and :hi. :he uses ויקיטיול, while :hi uses विकिट्रैवल. --Peter Talk 05:59, 23 August 2012 (UTC)
- Ryan, I will do the sed stuff. As soon as I have got an overview about tonight's new discussions. -- Hansm 06:39, 23 August 2012 (UTC)
- I don't understand the patterns with "Wikitravel Shared" and "Wikitravel Shared Talk". What do you thing where this could be used? -- Hansm 08:16, 23 August 2012 (UTC)
- "Wikitravel Shared" and "Wikitravel Shared talk" are the project namespace on shared: - see for example [11]. -- Ryan • talk • 14:56, 23 August 2012 (UTC)
This is going to become really technical. Please let's continue this on tech:Patterns to replace before the Import#en -- Hansm 12:03, 23 August 2012 (UTC)
Post-Import Scrubbing
editHere's the list of patterns I have right now (I've simplified the regex for discussion purposes, but for the technical folks whitespace and such will be handled):
- [[:(extra|openguides|wikevent|wikihow|wikioutdoors|world66):.]]
- Delete these - Obsolete interwiki links
- wikitravel.org URLs
- Update as appropriate - wikitravel.org/en/ would become en.wikivoyage.org, wikitravel.org/wiki/en/ would become en.wikivoyage.org/w/, etc.
- For questions, comments, or personal stories about this destination or topic, visit [http://extra.jamguides.com JAMGuides Extra].
- Delete - default text that appears on talk pages, no longer relevant
- wikitravel
- Replace with wikivoyage - will result in some cleanup being needed, but probably better to be overzealous rather than risk potential trademark issues.
- Википутешестви*
- Replace with Wikivoyage on Russian per Atsirlin
- википутешественник
- Replace with участник on Russian per Atsirlin
This would leave some references to Wikitravel Extra (although it will become Wikivoyage Extra) and WTP that may need some non-bot attention. Does that sound OK? Any other suggestions? -- Ryan • talk • 03:08, 21 August 2012 (UTC)
- Can we restrict these changes to non-talk namespaces? I'm worried that we'll end up with nonsensical conversations like "Ever since the Wikivoyage folks left Wikivoyage...", and talk page discussions should pose no copyright issues. There may also be old links on talk pages to extra or world66 that would cause issues if deleted outright. LtPowers 17:55, 22 August 2012 (UTC)
- That makes sense to me and should be doable unless anyone has any objections. -- Ryan • talk • 05:36, 23 August 2012 (UTC)
- Is wikitravel indeed a (registered) trademark anywhere? If so, can we get away with just noting in general terms and conditions "Wikitravel is a (tm) of ...". Straight search-and-replace will cause some very weird consequences, especially on eg. talk pages. --Kim Bruning 10:04, 23 August 2012 (UTC)
- I can't find it in the PTO's TESS database, so it's probably not registered. But unregistered trademarks have the exact same protections as registered ones, AFAIK. LtPowers 13:22, 23 August 2012 (UTC)
- Is wikitravel indeed a (registered) trademark anywhere? If so, can we get away with just noting in general terms and conditions "Wikitravel is a (tm) of ...". Straight search-and-replace will cause some very weird consequences, especially on eg. talk pages. --Kim Bruning 10:04, 23 August 2012 (UTC)
- That makes sense to me and should be doable unless anyone has any objections. -- Ryan • talk • 05:36, 23 August 2012 (UTC)
This is going to become really technical. Please let's continue this on tech:Patterns to replace after the Import#en -- Hansm 12:03, 23 August 2012 (UTC)
5. Technical challenges
editWT users and WV users
editEven if most WT users use the same name on WV, there is no way to make sure that the same user names on WT and WV identify the same physical person. I would prefere to treat WT user names and WV user names as logically different. Since the MW software is tolerant to unknown user names in XML dumps, I'd propose to do it the same way as we have done when importing de: and it:. This is to prepend the label "(WT/en)" to each user name in the XML dump file before importing it. Of course, this does not apply to IPs when contributions have been made anonymously.
- We can determine whether users are the same, if necessary, by having the user on Wikitravel declare that it is. --Peter Talk 14:55, 16 August 2012 (UTC)
- OK, in theory we can. But I do not have any idea how to manage and administrate this information. If I remember well, it is hard (to impossible) to change the user of a contribution afterwards in MW software. But this is what we had to do because most of the WT users would do their declaration after the import has been done, if at all. -- Hansm 16:39, 16 August 2012 (UTC)
- Users here can just state on their talk page that they are user X on WT and provide a link to the same. Jmh649 05:29, 17 August 2012 (UTC)
- OK, in theory we can. But I do not have any idea how to manage and administrate this information. If I remember well, it is hard (to impossible) to change the user of a contribution afterwards in MW software. But this is what we had to do because most of the WT users would do their declaration after the import has been done, if at all. -- Hansm 16:39, 16 August 2012 (UTC)
I just signed up at Wikivoyage and was happy that i can use the same account name as on WT. I think it would be terrific if we could reserve the names for the admins and some other established users. That would encourage some people to move with us. I see you use Chaptcha when creating an account but why i have to use it if i edit my user page? Sorry to be Off-topic with the last part. Jc8136 07:32, 17 August 2012 (UTC)
- Hi, Jc8136. For whatever it's worth, had you not asked your question about Capcha, I would have asked it, myself, so I tend not to consider it off-topic. (However, I don't seem to be getting a Capcha screen anymore.) Ikan Kekek 10:54, 17 August 2012 (UTC)
- Hansm has disabled Captcha for registered users now—it's only for registration. --Peter Talk 15:14, 17 August 2012 (UTC)
Pre-Registration of all WT user accounts
edit(Please, keep this discussion technical!!! No "it would be nice to have" contributions!) -- Hansm
Ryan and Jc8136 have proposed to pre-register all WT-accounts on WV and to set up a process where people could prove they are the corresponding Wikitravel users. This might be possible, but for sure not as easy as it looks at a first glance. We needed additional human resources for all the technical works. WV can not provide this. If Ryan or somebody else wants to, OK. I think it would not be a problem to grant Ryan SSH access to our server, if neccessary.
Here some technical and logical details:
Probably, we should automatically create a user page for each reserved user. This user page is read only and says that this user is a WT user and not yet registered on WV. After the WT account has been approved, the page must be toggled to be writable.
We can not be sure that the same user name on different WT wikis correspond to the same physical person. Thus, accounts can only be approved per wiki, not globaly. The consequence would be to keep the users of the migrated wikis separated from our global user database.
How could the proving process look like? Assumed, a new WV user wants to register for a reserved WT account. How to prove that he really is the owner of the WT account? Let's discuss the following solutions:
- Ask him for his WT password and verify it against WT using the API.
- Give him two tokens. He has to put the first one on his WT talk page, using a special template. Our system verifyes that the token is correct and the edit really has been done by the corresponding WT user. Again, this would require a working API on WT. If the proval has succeeded, the user can register on WV with the second token.
- add other practicable solutions!
Ways 1 and 2 require a working API on WT. If they toggle it off, there is no more way to proove the claim for a WT account. This means that all these old WT names will remain blocked for further use on WV, what might be an ugly situation.
-- Hansm 10:30, 17 August 2012 (UTC)
- There are two other options for confirming accounts. The first is simply the user confirming this, using their WT account—on WT. Regarding the second, please email me and we'll talk ;) The other basic way of handling this would just to be to assume good faith and deal with problems only if there is a complaint? We'd have to do that for already registered global accounts on Wikimedia, anyway. --Peter Talk 15:12, 17 August 2012 (UTC)
- If everyone is OK with Peter's third proposal (assume good faith and deal with conflicts as they come up) then I'm also fine with that. It's simple, and I would suspect there won't be a lot of cases where a user name is "taken" by someone else. -- Ryan • talk • 15:52, 17 August 2012 (UTC)
Importing
editImport Script
editA few questions for the Wikivoyage admins:
- Are we using importDump.php for the import? Currently the Wikitravel XML has been exported with one topic per file, so (for example) the China article is a single, 698MB XML file that contains all revisions for the Wikitravel China article (this is the largest XML file - most are less than 100kb).
- When you set up the German and Italian Wikivoyage sites, how did you import images? Looking at the process outlined at [12] I'm not sure that it will import images with revision history and uploader. Also, if we use the importImages.php script, I assume the file directory structure should be the same as it was on Wikitravel?
Any additional advice you can provide based on your past experience importing the German and Italian versions would be appreciated! -- Ryan • talk • 18:08, 18 August 2012 (UTC)
- Yes, we will use importDump.php.
- Unfortunately, I don't know any more. It's already 5 years ago that we have imported it:. I will have to dig my old import scripts, but not today, we already have 9:30pm. The only thing I can remember is that we had to do it in two steps, first the image and then the description page.
- -- Hansm 19:40, 18 August 2012 (UTC)
- At first the description pages, than images. In the first step there is of course no image. The second step is like changing a (non-existant) image. Of course, we will have only the last uploaded image, not the image succession. The description file history would be complete.
- Of course we have shell access to import. A 700MB file can be imported only on this way. --Roland 19:55, 18 August 2012 (UTC)
- I've uploaded all of the XML exports to the Wikivoyage servers and written a bash script to perform the import. I've done some tests on the "ja" instance that Hans set up and can verify that things seem to work - importing the 700MB China export went fine, imports of files in the MediaWiki: namespace worked, and importing of several hundred template files went fine. From the performance I saw I expect it will probably take close to a full day when we import everything. -- Ryan • talk • 06:01, 21 August 2012 (UTC)
- Does this imply importing full history? --Kim Bruning 10:05, 23 August 2012 (UTC)
- Yes. -- Hansm 10:19, 23 August 2012 (UTC)
What to Import
editIn an email today Hans asked about what to import and suggested "Maybe not user talk pages". I think he suggested this solely for discussion purposes, but I'd like to see everything from Wikitravel imported, including user talk pages, and if in the future we decide something is unnecessary it can then be deleted. For now, let's bring in everything so that we have it, rather than cherry picking and then realizing we should have imported something but chose not to. -- Ryan • talk • 03:13, 23 August 2012 (UTC)
This are all namespaces:
- MAIN, but not the main page
- Talk
- User
- User talk
- Wikitravel
- Wikitravel talk
- File
- File talk
- MediaWiki
- MediaWiki talk
- Template
- Template talk
- Help
- Help talk
- Category
- Category talk
I'd be also fine with importing "User Talk". But how about "Mediawiki" and "Mediawiki Talk". I'd assume, although not really sure, there is mostly Wikitravel specific content in this both ns. So, I'd tend to skip them in the import. -- Hansm 07:57, 23 August 2012 (UTC)
Is it OK for you to skip user talk pages of anonymous users, i.e. IPs? We do have thousands of them on en:. -- Hansm 09:46, 23 August 2012 (UTC)
- Aside from a slightly faster import and a slightly smaller database after import, is there an advantage to importing fewer pages? For example, IP user talk pages often contain messages to hotel spammers letting them know why their contributions are being removed. After import those same spammers will eventually find the new site and most likely still be using the same IP, so it would be nice not to have to re-notify each of them. As to the MediaWiki: namespace files, I would guess that perhaps half of them are Wikitravel-specific, so the question is whether it's better to exclude them or to import them and then update as needed - my personal preference would be to just import them, since it would be nice to have the history for pages like the site notice. -- Ryan • talk • 14:53, 23 August 2012 (UTC)
- @Unger - why not the main page? The import process can merge revisions, and while we'll modify it, I think we want to include that page in the import. -- Ryan • talk • 04:56, 24 August 2012 (UTC)
- To make it really anew and let it completely free from any kind of attribution to Wikitravel. We made it at WV/de and WT/it in a similar manner. --Roland 08:40, 24 August 2012 (UTC)
- @Unger - why not the main page? The import process can merge revisions, and while we'll modify it, I think we want to include that page in the import. -- Ryan • talk • 04:56, 24 August 2012 (UTC)
- I think redirects don't need an attribution and a history. Just importing the last version. And what about disamb pages? Actually the same. But we can not prove, that on these disamb pages ist not more content than a simple list. -- DerFussi 09:46, 25 August 2012 (UTC)
- Is there a specific reason not to import full history for these pages? In many cases a redirect or disambiguation page might have started off as a full article that was later merged, so the revision history might contain attribution information for content that was merged to another article. Similarly, the talk pages may show the discussion that led to a merge, and are thus useful for understanding why the merge occurred. -- Ryan • talk • 16:46, 25 August 2012 (UTC)
- To reduce the count of articles which have to be attributed and linked to Wikitravel. --Roland 17:26, 25 August 2012 (UTC)
- I think it would be safest to start with a full copy of Wikitravel. Since the import process is automated no extra work should be needed to import more articles, and for redirects and disambiguation pages the page size will be very small and thus should have minimal impact on the database and other resources. Put another way, if it isn't any extra work to import full history and talk pages now, it would be better to import them than to later find out that there is a reason that we should have imported them. -- Ryan • talk • 17:33, 25 August 2012 (UTC)
- I don't know any about the technical aspects here, but I'd really like to have the full revision histories of redirects imported, if at all possible. We had a template over at the old site, directing users to the merged & redirected page for attribution purposes. Please take a look, for example, at here (bottom of the page), to see the template in use. Vidimian 20:56, 25 August 2012 (UTC)
- I think that it would be problematic to skip old revisions of redirects. I don't know about Wikitravel, but sometimes users on Wikipedia move pages by copying and pasting (which is illegal: it breaks attribution). If illegal page moves are spotted, it is usually possible to find the old revisions in the history of a redirect. Once the old revisions have been located, the attribution issue can be fixed, but if you choose not to import the old revisions at all, you wouldn't be able to correct the attribution at all but would be stuck with an article which is a copyright violation (due to lack of a proper edit history). --Stefan2 11:38, 28 August 2012 (UTC)
Images
editBased on what I've found reading documentation there appear to be two viable ways to import images - a custom script that would use the Mediawiki API, or the importImages.php script that is provided in the Mediawiki installation. The pros and cons of each:
- A custom script that uses the Mediawiki API could be used to import an image with all revisions, but it does not appear that there would be any way to set the original uploader - all images would show up as uploaded from a single user, which might have CC-SA attribution ramifications.
- The importImages.php script can be used to import images. With this tool it appears that the original uploader can be credited properly, but only a single revision could be imported (once Hans grants me access I need to run a test to ensure that the importImages.php script cannot create a new revision if a file with the same name is imported, but it looks like it will not do so).
Note: the image description page is imported via XML, so that page will be imported like any other topic page with full history; the above options refer only to the image files themselves.
My preference is (I think) option #2, with the most recent image revision being the one that is imported, but the uploader of the first revision being credited for the upload, since they are most likely the original author. That strikes me as likely being the most accurate solution given the technical limitations. The loss of revisions is unfortunate, but the same thing would probably happen as images are moved to Commons, unless people planned to upload each revision sequentially when moving images to shared. Does this approach (#2) seem reasonable? Does anyone have any concerns or alternate suggestions? -- Ryan • talk • 06:01, 21 August 2012 (UTC)
- I do think that #1 would not be acceptable, as many license info sections have attribution to the uploader (e.g., PD-author). What would the #2 option look like? would a file like this [13] lack the File Info section? --Peter Talk 07:35, 21 August 2012 (UTC)
- If we go with option #2, [14] would include only the most recent revision (from 28 September 2010 02:28), but I would credit it with the author and comment from the original upload (Peterfitzgerald, "{{Imagecredit|credit=Peter Fitzgerald|captureDate=16 June 2007|location=Chicago|source=[[:Image:Bronzeville.svg]]|caption=Chicago's Bronzeville District|description=Street map}}"). -- Ryan • talk • 14:44, 21 August 2012 (UTC)
- I'm fine with the loss of revision history in #2. But I'm still a bit confused about how the credit will work... if I understand Ryan's comment above, I will be credited as the uploader of this image that I imported from Commons [15] and the original source and author information will be retained. Is this correct? -Shaund 15:50, 21 August 2012 (UTC)
- If there is a single revision, as in the image you cited, then nothing should be different from Wikitravel with option #2, except for possibly the upload date (I don't think the image importer allows me to set a date) - original uploader and comment will be maintained. It's only if there are multiple revisions that we have a problem, since the Mediawiki image import script does not appear to allow creation of multiple image revisions (to be verified, but that's my takeaway from the documentation). -- Ryan • talk • 16:01, 21 August 2012 (UTC)
- Would there be a way to preserve the history of revisions, despite not having the historical image files saved? I mean the text history of uploads at the bottom of the screen. I think there might be an attribution issue if that's lost, since plenty of people revise files without bothering/remembering to update the "author" section of the image credit template. --Peter Talk 18:04, 21 August 2012 (UTC)
- I'll need to do some testing. I think Hans was of the opinion that there might be a way to create empty revisions for attribution purposes, but I won't be able to test it out until after work tonight. -- Ryan • talk • 18:17, 21 August 2012 (UTC)
- Hans' script for importing images with revision history works, so the above concerns are moot - image information will be fully imported and attributed properly. -- Ryan • talk • 07:30, 22 August 2012 (UTC)
6. Schedule
editDoes the following look reasonable? I'm making a best guess since availability of technical resources may be limited - I'm on vacation from 26-August through 1-September, so will not be able to do much during that time:
- 15-August through 22-August: Open discussion of migration, investigation into technical issues, develop a migration plan.
- 22-August through 25-August: Migration dry-run. Attempt to load the data and verify that everything has migrated successfully. Any edits during this time may be lost if a reload is required.
This strikes me as somewhat aggressive, but since I'm unfortunately a bottleneck things either need to happen before the 26th or after the 1st. -- Ryan • talk • 18:14, 17 August 2012 (UTC)
- Ryan, basically, this looks reasonable, although maybe somewhat tough. First of all, we need to setup a new server. The current one urgently needs a system upgrade and there is an instability in the database, i.e. not a good bet for migrating lots of data. If you have skills in administrating a Debian/Linux system, your help is welcome. -- Hansm 10:14, 18 August 2012 (UTC)
- My background is fifteen years in web development (mostly Java) so I'm not an expert in system administration, but please email me with any specific questions and I might be able to help out. Would it also be useful to set up a new page on this wiki where we could outline any technical issues, and solicit feedback? I'm not sure if that would be appropriate or not since it's often advantageous to not publicize things that could later be used to attack the site, although in this case it sounds like most details would be fairly standard system setup issues. -- Ryan • talk • 17:51, 18 August 2012 (UTC)
- I'm not the bestestest sysadmin on the planet (I prefer dev and community work somewhat), but I try to keep current. Can I help out? --Kim Bruning 16:40, 19 August 2012 (UTC)
- I have a great deal of experience with Debian administration in general, but not much for database administration in specific. What sort of tasks could you use a hand with? -- D. Guillaume 05:00, 20 August 2012 (UTC)
- Kim & Dguillaume - has Hans responded to you yet? I know he was busy with several technical issues and unfortunately I don't know the specific sysadmin issue he needed help with, but if you haven't heard anything it would be great if you could email him to see if he still needs assistance. -- Ryan • talk • 23:49, 20 August 2012 (UTC)
- We need to get this up as soon as possible. If this story where to hit the NYTs we want someplace for people who are interested to go and begin helping out. Jmh649 04:59, 22 August 2012 (UTC)
- Kim & Dguillaume - has Hans responded to you yet? I know he was busy with several technical issues and unfortunately I don't know the specific sysadmin issue he needed help with, but if you haven't heard anything it would be great if you could email him to see if he still needs assistance. -- Ryan • talk • 23:49, 20 August 2012 (UTC)
- See #Progress below for the latest updates. -- Ryan • talk • 15:28, 22 August 2012 (UTC)
Images that need editing before migration
edit1. I have created a number of maps for WT which have a WT logo on them. Some of the source svg's were too big to upload, so are not on Shared. I intend to provide new versions without the logo, but I am not sure how best to go about this. Should I:
- Leave it until the basic migration has been done and the new site is up and running, then upload the new versions to Commons,
- Modify the files now and upload them to WV somewhere now (possibly a place specifically for this purpose),
- Modify the files now and upload them to Commons now, make any necessary corrections when the text is up and running,
- Do something else.
2. Will we be using logos on the new site? Obviously we don't have one yet, as the name hasn't been settled, but I don't remember seeing any suggestions for a policy on logos on maps. Cheers, Pbsouthwood 09:15, 18 August 2012 (UTC)
- Maybe I am wrong, but WT logo does not mean copyright? Since many of the maps need a logo, I would rather update them after a new logo is established. Otherwise, it is double effort: first go through all maps and remove the old logo, then put the new one. Atsirlin 12:48, 18 August 2012 (UTC)
- I don't think changing logos is critical as IB does not actually own the copyright on the logo - it was created by Mark Jaroski ([16]). As a result I don't think there are any legal issues surrounding it, and changing logos can probably wait until such details are determined for the new site. -- Ryan • talk • 17:14, 18 August 2012 (UTC)
- The triangle logo I agree should be fine, but the trademark "Wikitravel" is IB's and that could cause legal issues. LtPowers 02:20, 21 August 2012 (UTC)
- Partly relevant, I believe that eventually, not only should the logo be replaced in our maps, but possibly some of the styling. The style we have right now with roads, etc. is great. But things like the compass, key and location header should change to reflect a new logo and colour scheme. While we have always gone for the traditional turquoise/blue scheme, maybe it's time for a change to reflect a new site and a new dawn. JamesA 10:08, 21 August 2012 (UTC)
- Lots of work, and it will probably take quite a while to come up with an acceptable substitute, so for starters I think just dealing with possible trademark issues is enough. Everything uploaded to WT shared has to be CC-by-sa, and IB has been accepting all the maps with the WT logo on them under those conditions. Is it possible for them to try to claim exclusive rights on something they have condoned for years as free? Even in USA? Pbsouthwood 12:31, 21 August 2012 (UTC)
- Perhaps it makes the decision easier if you first look at our WV-Category-page Shared:Images by Source. We had no problems with this pictures, and until six months ago, we also used a few images with the WT logo. -- Berthold 13:03, 21 August 2012 (UTC)
- Peter, anything's possible. Even the flimsiest case, one destined to be thrown out of court if it ever got in front of a judge, could cause major headaches in the interim. LtPowers 14:42, 21 August 2012 (UTC)
- Give me a sign, if you need some help later. I can not help with the database stuff during the migration. But I am familiar with inkscape.... -- DerFussi 15:08, 21 August 2012 (UTC)
- The maps are licenced under CC-by-sa, so I'd think we should be fine, but as Lt Powers said, anything's possible.
- That said, Peter (Fitzgerald) and I were working on scrubbing the maps earlier -- Peter was working through the Africa maps and I was doing Europe. We both put our efforts on hold a little while back. I haven't restarted yet, not sure if Peter has. -Shaund 16:01, 21 August 2012 (UTC)
- I am happy to help scrub reference to Wikitravel from Australian maps, if needbe. But are we uploading them to a new site somewhere, or keeping them on our hard drive until the new site is ready? JamesA 04:35, 22 August 2012 (UTC)
- I will clean up all the dive site maps that I drew as I have the svgs, and will update them on WV as soon as possible. I can give a hand with others too if somone sends me a list. Pbsouthwood 15:32, 22 August 2012 (UTC)
Progress
edit18-Aug thru 21-Aug
editThe board as assured that the migration of Wikitravel content will be launched as soon as we can be sure that there is a significant support from active Wikitravellers. See the poll. In order to save time, technical preparations will start right now. -- Hansm 09:58, 18 August 2012 (UTC)
Wikivoyage has got new server hardware for hosting the new wikis. During the next days, it needs to be set up. -- Hansm 09:58, 18 August 2012 (UTC)
The basic system, the firewall and the mail server seem to work on the new wserver. Next steps will be to set up the Apache web server with PHP, the PostgreSQL database and a standard MediaWiki test version. -- Hansm 19:44, 18 August 2012 (UTC)
There is a MediaWiki test installation running, now. However, everything remains protected by a password until the import is done. I have decided to use MySQL instead of PostgreSQL, not because I love it, but because MW is too MySQL oriented. Next step will be a test import. -- Hansm 17:04, 19 August 2012 (UTC)
Finally, all Wikitravel content backuped by Ryan has been transferred to our server. All in all it was 91GB of data. Ryan and I have figgured out how to import all article revisions and revisions of files. The import scripts are prepared for the big run, but first some minor issues have to be clearyfied. We estimate that the actual data import will take more than two days, maybe even more than three.
Next steps would be to adapt our basic extensions running on de:, it: and shared: to the current version of MediaWiki software. We also have to find a way how to manage the data import for the extension that credits the original authors, the source work, etc.
-- Hansm 12:06, 22 August 2012 (UTC)
- I don't have much to add to what Hans has said above. As noted, the scripts to import XML and images are prepared, although as they have not been used for a full import before it is entirely possible an issue might be encountered during the import with a bad special character or some other issue. Assuming everything works it is expected that a full import will take at least 1-2 days. After the import is done we'll need help verifying that content has been properly imported (particularly for non-English versions). I'm also planning on writing a bot that will, post-import, go through each article and scrub data as outlined at #Scrubbing Data. Anyone with experience in running a Wikimedia installation is welcome to email me if you would like to be more involved, particularly if you have experience with more obscure aspects of Mediawiki such as OpenID integration or shared data repositories. -- Ryan • talk • 15:28, 22 August 2012 (UTC)
Status as of 22-August
editThe schedule for importing data originally listed today as the goal for starting imports of Wikitravel data, but unfortunately that is going to slip a bit. Here's an update on where things stand:
- All of the Wikitravel backup data has been transferred to the Wikivoyage servers.
- Import scripts have been written to perform the full import of topics and images. Topics will be imported with full history using the Mediawiki importDump.php script, and Hans has created a custom version of the Mediawiki importImages.php script that will allow us to import images with full revision history.
- Hans has set up several test instances of Mediawiki that have been used to verify the import scripts, and also ensure that some of the required plugins are configured and working.
These items remain outstanding and could use feedback from those who can provide it (list taken from an email Hans sent me today):
- Prefix for user names or not - see #Attribution and #WT users and WV users.
- What to edit before the import of the history - see #Pre-Import Scrubbing.
- What to import - see #What to Import
- What database to use (Wikivoyage currently uses Postgres, but Mediawiki generally uses MySql)
Hans' thoughts on these issues (quoted directly from email):
- As you have mentioned, at the end it's the association that has to take the resposibility for this.
- I think we should at least adapt the name spaces "Wikitravel:" and "Wikitravel Talk:" into "Wikivoyage:" and "Wikivoyage Talk:" in order to keep the links intact. However, the visible Text should remain as it is. When we prefix the user names, we also should adapt the user names on talk pages including the links to the user contributions special page.
- Maybe not user talk pages.
- I still don't know.
With the exception of #4 I think all of these items are being discussed elsewhere on this page, so in an effort to keep discussions in one place please continue discussing them in the linked threads.
Hopefully that's useful for those that might wonder what's going on. I don't have a lot of experience with Mediawiki configuration but will be working on a bot that we can use for post-import data scrubbing. If anyone wants to help out and you have extensive experience configuring Mediawiki, deep sysadmin skills, or strong MySQL DBA experience, please post here or contact me directly and I can put you in touch with Hans via email to see if he can use the assistance. Additionally, if you're a lawyer and can provide a legal opinion to help us ensure we are 100% in compliance with all terms of the CC-SA (see #1 above, for example) that would also be helpful! -- Ryan • talk • 03:12, 23 August 2012 (UTC)
Status as of 24-August
editAn update from Hans:
- I have set up shared2.wikivoyage.org for a first test import. Please don't make any further test with shared2 (and not even with en.wikivoyage.org) so that I can see what the scripts have produced.
The import of the Wikitravel Shared data is massive and will likely take a long time to complete, and that assumes that everything works on the first try. -- Ryan • talk • 15:45, 24 August 2012 (UTC)
Status as of 25-August
editHans completed an import of all Wikitravel Shared data (topics and files/images, including full history) to wts.wikivoyage.org last night. Note, however, that this import does not include the data scrubbing tasks mentioned elsewhere on this page, but it is a good test to prove that the import process works. It is entirely possible that this first version will be deleted and re-loaded once scripts for the data scrub have been put in place. At the present time the test site remains password protected since there is a danger of having it publicly available while setup is ongoing - there is enough to debug without having to wonder if a problem is due to someone changing a page, etc. -- Ryan • talk • 18:26, 25 August 2012 (UTC)
Status as of 26-August
editA test import of Wikitravel en: has been completed, and so far, things look good. I have imported all images and the current revisions of all articles (from all namespaces). en: and wts: (the former Wikitravel/shared) work well together.
Next step will be to do post-import scrubbing and attribution.
-- Hansm 10:31, 26 August 2012 (UTC)
Status as of 28-August
editI am aware of the dispute about prefixing WT user names with something like WT-en. We have asked for legal advise, but things don't seem as clear as we wanted to have them. There is no definit answer until now. Anyhow, I have started the full data import for en: about 24 houres ago. So far, we are at "S" in the main namespace. Also, there are tested scripts for insterting the attribution data into the database and I see a resonable way how to display them as demanded by our board. Excepted the outstanding dispute about user names, there is only one more thing to find out: How will we keep our existing single sign on feature when we have two servers running with different MW versions and no LAN between them? And, of course, all wikis on differnt sub-domains. -- Hansm 20:55, 28 August 2012 (UTC)
- I am not that familiar with that technical stuff and can not help, sorry. Is there a solution at all? Can we use the WMF solution? If both answers are: no. We have to live with that. I have no problem to login twice. Can we copy our user table to the new wikis? Within the next few days we will get a comment form the WMF, I think. If we are heading to a WMF project they we can ask the WMF staff in preparation of the next transfer. -- DerFussi 06:30, 29 August 2012 (UTC)
Status as of 29-August
editThe article import has completed right now. Next is to import images. -- Hansm 09:19, 29 August 2012 (UTC)
Stopped the attempt to make an intermediate zipped db backup after about an hour as it has turned out to be far too slow. Unzipped backup started. Just in case anybody is wondering what I'm doing all the time. -- Hansm 10:42, 29 August 2012 (UTC)
Images have been imported to new en: and attribution is done. Next will be post-import scrubbing. Still not Idea how to manage a common user database for all wikis. -- Hansm 15:45, 29 August 2012 (UTC)
The import of post-import scrubbed articels is running on en:. Will complete in about one hour. -- Hansm 20:05, 29 August 2012 (UTC)
Status as of 30-August
editUnfortunately, there still have been some bugs in the post import scrubbing script. I only did realize after already having ruined the database. Thus, I had to replay the backup made after the first article import. At minimum half a day got lost. -- Hansm 10:48, 30 August 2012 (UTC)
en: is migrated. Roland is working on adapting some extensions in order to make everything looking as it used to be. Now, since my migration scripts are (hopefully) elaborated enough, I hope wts: (WT shared) will go much faster. Except ru:, I have not yet got any patterns for adaptions of other languge versions. Please submit to tech:Patterns to replace after the Import. -- Hansm 19:43, 30 August 2012 (UTC)
- That's great to hear. Do you have an idea of when we will have :en up and editable? I'm getting a lot of emails asking that question. --Peter Talk 20:15, 30 August 2012 (UTC)
Status as of 2-September
editI'm sorry for the delay and the lag of information. I have been sick for three days and still I am. There was only few things to do, but unfortunately, I was unable to do. Now, together with Rolands help, en: and wts: (WikiTravel Shared) are up for a testing period of 2 days. You can edit as usual, but keep a local copy of important edits since they still might got lost due to bugs! So far, Peter and Ryan are bureaucrats.
Please report bugs on tech:Wikitravel Migration bugs.
The wikis are on http://en.wikivoyage.org/wiki/ and http://wts.wikivoyage.org/wiki/ , still protected by a password. User: tester, password: test.
-- Hansm 18:32, 2 September 2012 (UTC)
UPDATE: Unfortunately, I had to disable this tester access. -- Hansm 18:37, 3 September 2012 (UTC)
- Feel better soon! Ikan Kekek 19:26, 2 September 2012 (UTC)
- Incredible! The new wikis look very good. Their bugs are mostly inherited from WT. Thank you very much! Atsirlin 19:44, 2 September 2012 (UTC)
- Yes, feel better soon and thanks for doing this! -Shaund 19:59, 2 September 2012 (UTC)
- Well done, It seems to work and looks right. I updated a map and it loaded and displayed consistently. There does seem to be a small aspect ratio anomaly on some image files which are slightly flattened in en: but show correctly in WT-shared. I wish you a quick recovery. Cheers, • • • Peter (Southwood) (talk): 20:58, 2 September 2012 (UTC)
- Thank you for your hard work! Unfortunately, I got the "Error 504 Gateway timeout" message every time I tried to view either site (around 0:30-0:45 UTC Sept. 3). AHeneen 00:46, 3 September 2012 (UTC)
- This is great! Thanks very much for your work, and get well soon. JamesA 06:42, 3 September 2012 (UTC)
- There seems to be a problem in the New York City guide[17]. The top line is showing up with overwriting in my browser, Firefox 15.0 in a Windows XP environment. I'm also getting a login error. Am I the only one whose password is not being accepted? Ikan Kekek 07:42, 3 September 2012 (UTC)
- I don't see any issues on the New York article, nor have any login problems on Chrome for Windows 7. However, while browsing earlier, I occasionally found issues with the breadcrumb navigation, where the ">" would be replaced by weird script. It wasn't localised to specific pages, but just appeared randomly on any page then disappeared when refreshed. JamesA 11:48, 3 September 2012 (UTC)
- Ikan, do you mean that the coordinate-info to the right overwrites the localization-info to the left, just below the line at the title. It seems that coordinate-info behaves that way everywhere, even on wikipedia. (seems to work a bit better on german wikivoayage though. I will add it to the "bugs".
- And regarding the login, tI have some "problems" too, but it works anyway. Do you also get several login pop-up-windows? I just fill in the first and close the others./Johan Jönsson 14:01, 3 September 2012 (UTC)
- Johan, yes, the problem was that the latitude/longitude info was overwriting the right side of "Earth > North America > United States of America > Mid-Atlantic > New York (state) > Metro New York > New York City." That's not happening now. The problem with logging in is I tried to log in as Ikan Kekek to attempt to edit the article and was unable to. It thought I was typing in a wrong password all the time. I tried it again just now, and it still thinks my password is incorrect. Is it not really editable right now? Ikan Kekek 17:48, 3 September 2012 (UTC)
- I don't see any issues on the New York article, nor have any login problems on Chrome for Windows 7. However, while browsing earlier, I occasionally found issues with the breadcrumb navigation, where the ">" would be replaced by weird script. It wasn't localised to specific pages, but just appeared randomly on any page then disappeared when refreshed. JamesA 11:48, 3 September 2012 (UTC)
- There seems to be a problem in the New York City guide[17]. The top line is showing up with overwriting in my browser, Firefox 15.0 in a Windows XP environment. I'm also getting a login error. Am I the only one whose password is not being accepted? Ikan Kekek 07:42, 3 September 2012 (UTC)
Thank you all for the good wishes. It looks like it has helped. Slowly, I feel better. -- Hansm 14:18, 6 September 2012 (UTC)
Status as of 5-September
editBOMB! The error on images with multiple versions has a rather tricky reason. It is caused by the way the mediawiki software archives older versions. More precisely, the importImage.php maintenance script shipped with MediaWiki fails when it has to import different versions of the same image within one second. I could fix this bug by modifying the script. But unfortunately, now, I have to re-import all images with multiple versions on both, en: and wts:. Even more unfortunately, it's rather hard to extract the images that need to be re-imported. It would be much easier for me simply to re-import all images on both wikis. -- Hansm 18:04, 5 September 2012 (UTC)
Status as of 6-September
editToday's big news: The Wikimedia Foundation Board says yes! This means that work to create the site on Wikimedia servers is about to start. See full text of announcement. Jpatokal 11:13, 6 September 2012 (UTC)
I have to rebuild en: and wts: from the database dumps taken at 2-September in order to fix the problem with multiple file versions. wts: is done and should be OK, now. en: is in progress. Since there have been many edits during the testing period, I have saved the en: and wts: test wikis on http://en2.wikivoyage.org/wiki/ and http://wts2.wikivoyage.org/wiki/, respectively. They will remain read-only.
I think we should be patient enough to run one more test before releasing the new en: and wts: into productive use. The new testing period will go until Saturday, 2012-09-08, 10:00 am (UTC). If there won't be any bugs, we can go productive. Again, bug reports go on tech:Wikitravel Migration bugs.
-- Hansm 10:35, 6 September 2012 (UTC)
En: and wts: are up for testing, again. All template related bugs can easily be fixed by copying Roland's previously made fixes from en2: or wts2: over to en: or wts:, respectively. -- Hansm 14:17, 6 September 2012 (UTC)
- I assume we should still be keeping back ups of all articles to be updated, as we are still in the testing phase? I am eager to import all the non-backed up content that I am storing ;) --Peter Talk 19:22, 6 September 2012 (UTC)
- Yes. It's still a testing phase and if things go wrong, all edits may get lost again. So I urgently recommend to keep local backups. -- Hansm 20:42, 6 September 2012 (UTC)
- In cases where we have made edits to what is now en2: or wts2:, how should those changes be transferred to the later versions? Should the corrected files be moved or copied? • • • Peter (Southwood) (talk): 06:07, 7 September 2012 (UTC)
- Since no-one has answered my question I am going to copy. • • • Peter (Southwood) (talk): 16:45, 7 September 2012 (UTC)
- In cases where we have made edits to what is now en2: or wts2:, how should those changes be transferred to the later versions? Should the corrected files be moved or copied? • • • Peter (Southwood) (talk): 06:07, 7 September 2012 (UTC)
- I would say just copy the current revision of the modifications made on en2: or wts2:. A propper export-import-procedure seems unnecessarily complicated to me. -- Hansm 17:16, 7 September 2012 (UTC)
Status as of 7-September
editHaha, now, you all try it the lazy way and wait until the testing phase is over. This does not work! Do not expect devellopers or admins to find their own bugs. It is important that you test the current installation! Anyway, you will, either now or after the release of the wikis. So, better now. And, please, post bug reports to tech:Wikitravel Migration bugs. -- Hansm 13:49, 7 September 2012 (UTC)
- Maybe we haven't found any?? :-) I checked most of the stuff I reported last time and it seems OK now, you've done a good job. I found something small this morning so I'll be reporting it shortly. -Shaund 14:25, 7 September 2012 (UTC)
- Sometimes I have other work, but I will try to fit in a couple of hours tonight. I am off line for the weekend. • • • Peter (Southwood) (talk): 16:42, 7 September 2012 (UTC)
- I am unable to access it, is the user/password not tester/test anymmore? –sumone10154(talk) 19:43, 7 September 2012 (UTC)
- No, there is a new username/password you can get from Peter Fitzgerald by email. • • • Peter (Southwood) (talk): 20:26, 7 September 2012 (UTC)
- I haven't been able to access either site yet. Out of about 8-10 attempts to view it the past few days, only once did I get to the screen to enter user/password. It shows it's loading for a couple minutes, then I get a "HTTP Error: 504 Gateway Timeout RequestURI=http://en2.wikivoyage.org/wiki/". AHeneen 21:05, 7 September 2012 (UTC)
Status as of 8-September
editI have looked carefully through the bug reports. Most bugs are still unsolved, but I'm sure that we can fix all of them while normal operations are going on.
In other words: en: and wts: are released for productive work.
We will still keep them protected by a htaccess-Password until all WT logos are scrubbed. If you don't know the password, send a mail to Peterfitzgerald.
-- Hansm 13:37, 8 September 2012 (UTC)
Status as of 9-September
editru: is up for testing. I'm waiting for more language specific replace patterns and verification of my geusses for the names of the namespaces. See the corresponding language on Interest in starting a new language version for more information. -- Hansm 10:34, 9 September 2012 (UTC)
NPOV and No Original Research?!?
editWhat was decided about NPOV and No Original Research?
Wikimedia is not so taliban about opinions and user feedback, so do we just stick to our "play fair" policy?
After all, the act of choosing (in a big City) just a few hotels, restaurants and sights is necessarily subjective. I don't think we should be too rigorous about contributors opinions - that's what makes Lonely Planet so interesting and useful!
To me it is obvious that No Original Research can not be adopted for wikivoyage - I need to know if that bridge has collapsed or prices gone up drastically when I travel and who better than someone who has been there to tell me!
(Also needed to retain and encourage editors!) W. Frank 11:44, 19 August 2012 (UTC)
- There is a discussion of NPOV (and similiar Wikipedia policies) at [18]. The quick summary is that those are Wikipedia policies, and not Wikimedia policies, so a new Wikimedia travel site would not be required to adopt them. As you and others have pointed out, while they are good policies for an encyclopedia they would be detrimental for a travel guide. -- Ryan • talk • 16:19, 19 August 2012 (UTC)
Whew! I'm so glad if that is indeed the case, Ryan. Otherwise it would be a case of out of the frying pan and into the fire... W. Frank 16:53, 19 August 2012 (UTC)
- Yeah, Wikimedia's current most famous project does sort of cast a shadow, doesn't it? Wikimedia is more than just Wikipedia! (See http://www.wikimedia.org/) .
- NOR is a requirement for wikipedia; other wikimedia projects such as Wikinews explicitly do original research.
- NPOV is a form to ensure objectivity that is used on several wikimedia projects, but is not required on projects where it is not relevant.
- Is there a document on "fair play" I can point people to? :-)
- --Kim Bruning 16:50, 19 August 2012 (UTC) You think maybe with some more TLC, the travel wiki could become as famous as the wikipedias? O;-)
- Yes, Wikitravel:Wikitravel:Be fair (Wikitravel:Wikitravel:Business listings reliability Expedition is another interesting one). The talk pages of both of those are also essential reading. The discussions on the Meta proposal [19], especially the subsection Fair are pretty useful.
- Above all, though, I wouldn't worry to much about this point. The starting point for the project will be current policies (which are occasionally different across language versions), and the "NPOV pushers" will probably head back to Wikipedia, or realize that an NPOV travel guide is not desirable, after doing some actual contributing. Think about it: an NPOV travel guide would be required to represent the opinions of touts, scammers, aggressive advertisers, etc.!
- As a primary source, there just is no practical way that NOR could be applied, nor would it be desirable. These complaints, I think, will fade when people become more acquainted with what a community-sourced travel guide actually means. --Peter Talk 17:00, 19 August 2012 (UTC)
Policy organization clean-up
editThis section has been moved to Talk:WMF Migration/Policy clean-up, with the "New policies" discussion moved to Talk:WMF Migration/New policies.'
WM travel guide site setup
editDoes anyone know what functionality we can expect from the new site? Will it be set up the same as English Wikipedia for things like templates etc. so they can be copied over and still work? • • • Peter (Southwood) (talk): 17:12, 24 August 2012 (UTC)
- I think we will have a similar set of extensions including our own ones. Almost all of Wikipedia templates should work but they must be copied. As I learned we can wish which extensions are needed. And maybe we can install them by ourselves. --Roland 17:26, 24 August 2012 (UTC)
Move discussions to new pages
editThere's a lot of topics covered in the discussions on this page. Is it possible to move sections to new pages like WMF Migration/Shared, WMF Migration/Policy clean-up, WMF Migration/New policies, WMF Migration/New features and then move the relevant sections on this page to the appropriate talk page? I think creating the new pages would also make it clear what needs to be done. AHeneen 00:32, 27 August 2012 (UTC)
- I agree. • • • Peter (Southwood) (talk): 06:43, 27 August 2012 (UTC)
- New page created for WMF Migration/Policy clean-up per AHeneen above. • • • Peter (Southwood) (talk): 12:38, 28 August 2012 (UTC)
- I have gone ahead and moved policy clean-up discussion to Talk:WMF Migration/Policy clean-up. AHeneen 01:46, 3 September 2012 (UTC)
Expeditions
editI noticed the list of expeditions on WMF Migration/Policy clean-up. Is it appropriate to import all the expeditions from Wikitravel as-is? The issue I see is a lack of users taking part in each expedition, as some haven't been active in a while. Bringing them to WV where no one or only a couple people join may leave them as dead weight and if the need arises, a new one can be formed at a future time. It's probably best to start new expeditions, brought up at the lounge (is there any trademark issues with renaming it the "Travellers' Pub"?) by WV users. Of course, there's a couple good ones worth carrying over like Country surgeon & UNESCO World Heritage Sites. Plus, I think we'll get a lot of new users if/when we join WMF. In that case, it's probably best to start new expeditions and maybe have a list of the expeditions on WT somewhere to reference for ideas. Also, I think some of the pages we're working on now (like policy clean-up) should be transitioned to expeditions once all the English WT content is brought on-line here at WV. New language expeditions would be a part of Interest in starting a new language version and then, when created, expeditions to write policies and such would be on the new languages name space (fr.wikivoyage.org, ru.wikivoyage.org, etc.). AHeneen 07:53, 1 September 2012 (UTC)
- I think we should keep the expeditions which had a useful result or are still active/potentially active. I don't think they take much space. Those which fizzled out? I don't have a strong opinion either way. Mark them with keep or delete and see what the consensus comes to?
- I think we can keep the traveller's pub if we want to.
- Agree on the Migration Expedition. • • • Peter (Southwood) (talk): 08:38, 1 September 2012 (UTC)
- In the past we have only frozen expeditions and archived them, rather than deleted them. Our most successful expeditions (off the top of my head--I'm probably forgetting some) have been the regions map expedition, mapmaking expedition, and routebox expedition. There's little harm in keeping dormant expeditions around, though, since someone may find them and decide to give them life anew--just as long as we have consensus behind the goals of the expedition.
- I don't know if the expeditions should be part of the policy hierarchy though. Maybe just links in the "see also" sections at the bottom of policy articles? --Peter Talk 15:07, 1 September 2012 (UTC)
- The intention was for Expeditions as a concept/tool/way we work to be mentioned in the policy, but all the actual expeditions to be linkes as related pages, which is equivalent to see also, for easy reference. Do you think it should be different? It is in "Other" at present, because I couldn't think of a better place at the time. • • • Peter (Southwood) (talk): 16:38, 1 September 2012 (UTC)
- I don't have a strong opinion either way, but I think some of them have matured enough that they could be turned into policies or MoS guidelines -- like the Climate expedition and Routebox navigation. In terms of where Projects fit in the hierarchy, I think they should be outside of it, but a policy or guideline would reference a relevant project (e.g., Article templates/Sections should reference the Country surgeon expedition). -Shaund 16:50, 1 September 2012 (UTC)
- Routebox navigation is already in the manual of style.
- Should we move discussion of Climate Expedition outcomes to the new policies discussion page? I have put a placeholder link to Climate Expedition in the "Structure of travel guide articles" section.
- Your suggestion on Projects (Expeditions?) looks appropriate to me. Do you have any opinions on where it should go? • • • Peter (Southwood) (talk): 18:22, 1 September 2012 (UTC)
- Since Wikivoyage has the Interest in starting a new language version page, do we still need all the separate pages for Language Expeditions? –sumone10154(talk) 23:17, 9 September 2012 (UTC)
- This is one of many (I think) slight policy differences to be resolved between WV:General and WFKAWT:WTS (wiki formerly known as). I have added this bit to the beginning of a list here. --Peter Talk 04:01, 10 September 2012 (UTC)
Welcome messages
editThe "welcome message" template should probably be updated to help new users understand what's going on around here (there's no en:Wikivoyage yet & we're in the middle of the WT>WV>WMF transition). I'd do it myself, but can't find the template page when looking around the "Help" category or using advanced search for Templates or Wikivoyage namespaces. New users might be a bit lost trying to understand the site without guide pages or help topics...especially if they miss the banner message (I realize it's highlighted). Of course, the message can be updated once the WT content is loaded and live for anyone to contribute. AHeneen 05:37, 9 September 2012 (UTC)
- You need to modify Template:Welcome. Tell the people that there are en: and shared (called wts:) running on Wikivoyage, but not yet accessible to public since there are still Wikitravel logos that need to be scrubbed, first. Also, tell them that they should send a mail to one of tthe three listed in the launche section of the en language version intereset. -- Hansm 09:42, 9 September 2012 (UTC)
Imported user pages
editHans seems to be of the opinion that we have to keep our imported user pages (e.g., en:User:(WT-en) LtPowers) "frozen" and has reverted several of us who moved those user pages to our simpler Wikivoyage usernames. Can anyone think of a reason why we need to keep the (WT-en) user pages in a frozen state? LtPowers 19:00, 10 September 2012 (UTC)
- We have had long discussions and differnt oppinions about how to exactly comply the CC-by-sa in practice. See Talk:Migration FAQ#Legal issues and also Talk:Migration FAQ#WT users and WV users. We have tried to get legal advice from WMF lawyers, but even there we didn't get any help. At the end of the day, it is the Wikivoyage Association as the site owner which is in full legal resposibility for everything that happens on this site. And there is a simple judical rule: the party that has the resposibility also needs to have the right to decide. In other, very clear words: There are decisions to make that are not subject of community consensus. However, everybody is welcome to become member of the WVA and participate on decision making.
- The WVA, represented by its board, has given a statement on how to realize the import. Indeed, there is nothing said about how user pages have to be imported, but as the preferred way is to prepend the Wikitravel user name by a marker like "WT-en", it seemed logical to me that even the corresponding user pages need to have the same prefix.
- From the technical point of view, this is somewhat delicate as the prefixed user page does not belong to any wiki user on the new wiki. However, it is the point where people come to when they wanted to get more information on what this strange "WT-en" in front of some user names does mean. This is why I have put the info box at the top of this user pages.
- Honestly, I can not believe that anybody who has removed this boxes or moved the whole page did not get the idea that this somehow might have to deal with attribution info. Unfortunately, I did forget to protect this pages. I have done this, now.
- Attribution is simple and consists of the username in the history -- that's it. User pages are for collaboration, not for attribution. I even left the template you placed at the top of my user page when I moved it, just in case. If you don't let us redirect the (WT-en) pages to our "real" pages, how is anyone supposed to contact us about an old edit? LtPowers 21:03, 10 September 2012 (UTC)
- Personally, I assumed that as long as the wikis remain password-protected, we know who is who. Therefore, it would be natural to allow everyone to claim his/her identity, and block the remaining WT-en pages before the site goes public. But it is your prerogative to decide. Atsirlin 21:10, 10 September 2012 (UTC)
- Thanks to Atsirlin. As long as there is no reliable mechanism to prove the identity of a user, it is the safest way to strictly keep separated WT users and WV users. Maybe later, somebody is able to establish such a mechanism.
- @LtPowers: I know, many people do believe (or at least do propagate) that CC-by-sa attribution is as simple as just giving some user names. If this were true, the full legal code would be just a few sentences. It is not. And anyway, as long as we do not get any reliable legal advise, we rather will be overcautious.
- -- Hansm 21:32, 10 September 2012 (UTC)
- WT accounts cannot be claimed by new WV users, as we can never be sure if they are the same "person". This I understand. Completely separate from that discussion, is the editting of user pages. User pages are just wiki articles. Everyone can already edit everyone's user page, or any wiki article for that matter; that's the whole idea of a free wiki. You don't need to lock certain pages to comply with the CC-by-SA licence, you need to attribute which user made which edit. --Globe-trotter 22:53, 10 September 2012 (UTC)
- Hans, have you asked Creative Commons for advice on what the license requires? This is the first time I've heard that keeping a user page intact is a necessary part of attribution. LtPowers 23:10, 10 September 2012 (UTC)
- Agree totally with Globe-trotter. Two additional points
- 1. My copied userpage currently says in the first line that I've made thousands of edits to WV, when I haven't.
- 2. If I blank my userpage at WT, can the blank copy then be updated on WV? --Inas 23:18, 10 September 2012 (UTC)
- Okay, I've now blanked my userpage at WT. I've mirrored that change over to WV. I've created a new user page for myself on WV. No probs. --Inas 23:25, 10 September 2012 (UTC)
Identifying the same user on WT and WV
editI recently went through the process at Wikimedia of identifying myself as an account owner of another account on another WMF server. The process was simple. You log into your account on one WMF server. Make an edit saying "I am user:XXX on YYY server". You then undo that change. You then make a link to the item in the history, and post that on the initial server. Then you are accepted as the same user on both servers. Is there any reason we can't use a similar process here? --Inas 23:42, 10 September 2012 (UTC)
- I am not a professional programmer. Maybe its possible.... But there is code to be viewed and adapted... tests to be done.... Hans had to work a lot recently. Besides all this. Can this process wait till we are all moved to the WMF? They have everything - technichans, lawyers, and the technology running an their servers already.... The project is confirmed.... We should focus on bringing the wikis online. The community wants to work with it. And there is still a lot of work to do (removing the WT logos, migrating your policies .....) before we switch the wikis over to the public. -- DerFussi 05:40, 11 September 2012 (UTC)