Place new comments at the bottom

Add new comments, don't edit old ones

Parent Page > Discussion

Sign your comments using @SIG@

Edit conflict

This page is undergoing an edit war since a month. In March 2007 I did a cleanup of this page, but the changes were reverted by EddyPetrisor. I asked Eddy to justify his changes several times, and one day we came close to have a good discussion but stopped before significant progress was made. I came back with this on IRC on 2007-03-01 and told Eddy that I wanted the issue to be settled before April 2008. That week we did significant progress towards a merge, but not much. I failed to contact Eddy via IRC since then and was not contacted. On 2008-04-01, after waiting a year, I "hijacked" the page, but Eddy re-hijacked. So I'm trying another medium.

Eddy, first let me be clear: your behavior is unacceptable. Nevertheless, I will not take action if you do not push your version before mentioning a reasonable date before which you'll have time to justify your changes, and if you do justify your changes before that date. Feel free to justify on this page or via IRC. -- FilipusKlutiero 2008-04-02 01:33:34


One, the war has been going one for over a year and started with your refusal to understand that problems which appear in Debian or Ubuntu, because one certain distribution method is used, no matter where those problems originate, are problems (see your removals in commits 16, 17 and 18) and must not be hidden. -- EddyPetrisor

Two, the "cleanup" you talk about was mostly a replacement you tried to do in 20, but that didn't took into account that it was broken, since the "cleanup" contained a set of omissions that were killing valid information. -- EddyPetrisor

Three, your victim-playing and accusations have little, if any, grounds, and if you or anybody else wants proof, I still keep the IRC logs. Still, I will proceed here and enumerate the main issues with your "constructive" changes. -- EddyPetrisor

I have told you a loooot of times why your changes are unacceptable; while we didn't succeed to talk on he phone, I have told you on IRC, but you seem to be reluctant to accept the valid points I have made. I have explained you in more than one way why your changes are unacceptable, but you seem to think that omitting problems:

and information:

is the way to got forward. -- EddyPetrisor

Moreover, this page describes current way of distributing translations, and you obstinately remove sections, information and methods that are subject to this. For example, you remove the references to the issues that language packs (as iceweasel uses) have. With your last revert 39, you removed (or better said, remerged it in another) the section about the third type of distribution, the "one package for more apps (the KDE way)", while I have explained to you (since you don't seem to consider it warrants a different section, in spite of its obviousness to me) why this is a different thing. You still don't seem to understand that there is a difference (as proven by your manual revert 39 which replaces, among others, "There are currently several methods to distribute localized data" with "This document describes two currently used methods to distribute localized data...").

This alone should be a real sign of something being wrong since the old version actually did have three valid subsections, while your 39 version has only two and explicitly states that.

Also, note that I have been the one that actively merged some of your changes (http://wiki.debian.org/I18n/TranslationDataDistribution?action=diff&rev2=34&rev1=33), while you have reverted that merge blindly to your version. -- EddyPetrisor

Also, when you partly understood that there are other issues, after lengthy discussions with me on IRC, you understood that partly, but still did a bad job at merging what you understood - as proven by the version 31. -- EddyPetrisor

During our IRC discussion, you've made me loose my temper with your refusal to understand, so please stop playing the victim role and actually look at the changes you revert, you should know I have. Also, I don't appreciate your constant bashing and unfounded claims that you have waited for over a year, as you said in the commit message of version 35 ""hijack", after waiting for more than a year (still open if you can find time)", while the commits 26, 27, 28, 30, 31 prove - in spite of your obvious mistakes in understanding my points - that I have actively worked on fixing this issues. 2008-04-02 12:25:26 (with updates and corrections later) -- EddyPetrisor


I will revert YET ANOTHER TIME and warn you NOT TO REVERT YOURSELF AGAIN or I will be forced to appeal to and external arbiter or the tech-ctte. There are more things to add to your answer, but I won't do it now and I will address just the regressions you wanted to be pointed out to. Also, I will consider that, once I have shown you a lot more than your minimum required quantity of regressions, you have had accepted already that you were wrong, and any attempt to revert to your version will mean that you are ready and want to see the arbiter or the technical committee about this issue, before we conclude something out of this discussion. Here are more than what you requested wrt regressions:

1)

- Language packages are usually built from a special source package (e.g. iceweasel-l10n).

+ Language packages are built from the iceweasel-l10n source package. iceweasel doesn't depend on its language packages.

and iceweasel is definetly NOT the only appplication doing this, it is plainly a particular case. Generics are preferable since we're talking about methods of distributing localization info, particular cases should be just examples. The former wording is clearer and you revert introduces a regression.

2)

- This is the simplest solution. Localized data is included in the application package.

+ This is the simplest solution. Localized data is included in the localized package (or, rarely, in one of its dependencies).

This is non-sense on multiple accounts and another regression since the old wording was clearer.

  1. What the hell is the "localized package"? There is no definition of such a thing in the Debian land and that term is ambiguous. Is that a special package that contains strictly localization info or are you referring to the application package as "localized" once it has localization info in it?
  2. If the translation is included in the dependency, then, we're no longer talking about the first case (where the localized info is part of the main package).
  3. If there is a separate localization package, then why the hell is this part of the "translations are part of the main package"?

The proper way to word this is to say "This is the simplest solution. Localized data is included in the main (e.g. application) package." . (DO NOT TAKE THIS AS AN INVITATION TO DO THE EDIT YOURSELF since this was already addressed in the latter comments!).

If you still think that the case where the entire (i.e. all languages) localization info is distributed in a -data-like package which is a dependency (it is possible, although I suspect this is not, due to dependency issues, de-syncs that may appear), then that warrants another case or, at least, it should be clarified as such in that paragraph, which, unsurprisingly, WAS ALREADY DONE AGES AGO BY MYSELF:

- * Bundling localization data for all available languages in a binary package (either the application binary or a -data package). The main issue with this approach is the size of the package.

+ * Bundling localized data for all languages with the localized package. The main issue with this approach is the size of the package.

3)

- There are currently several methods to distribute localized data:

+ There are currently two methods to distribute localized data:

As you can plainly see in revision 38 (http://wiki.debian.org/I18n/TranslationDataDistribution?action=recall&rev=38) I was able to describe 3 (three) methods, this would mean, for any rational person the statement "There are currently two methods to distribute localized data" is FALSE.

More than that you yourself were able to come up with 4 (four) methods -- This creates 4 combinations -- as of which the last is not used since, ttbomk it would create all sorts of pain to the maintainer. THUS, we're left with 3 DIFFERENT VALID DISTRIBUTIONS METHODS THAT ARE PRESENTLY USED. Did that get through to you? Are we clear at least on this small detail? -- EddyPetrisor


People, this edit war and so-called "discussion" is more and more annoying. Philippe, this activity was started by Eddy and even if you disagree with his approach, please don't interfere with his work. You can start your own. -- ChristianPerrier 2008-04-05 14:54:39

Comments

Eddy, please keep my replies in their context. If you want to keep your whole replies vanilla, you can always duplicate them. -- FilipusKlutiero 2008-04-07 13:24:00