Re: D-I Manual - please update translations (update)
On Sat, Jul 22, 2006 at 06:24:09PM +0200, Frans Pop wrote:
> On Saturday 22 July 2006 11:17, Jens Seidel wrote:
> > On Sat, Jul 22, 2006 at 01:08:56AM +0200, Frans Pop wrote:
> > > Note that translations that are not sufficiently up-to-date will be
> > > excluded from the release.
> I feel this whole mail is a ridiculous over-exaggeration triggered by a
> different discussion that is currently going on on the debian-boot list,
> but I will answer it anyway.
No, I was indeed very horrified when I read your mail. There was
(initially) no influence from the debian-boot list. But you're right,
lets discuss this more seriously.
> > Once this is established for a beta release it will also apply to the
> > final version.
> In principle yes, but on the other hand I will, as a responsible release
> manager who cares about translations, allow a lot more time for updates
> and thus a longer freeze period.
I really don't care about time for updates and I'm also against any
freeze. It's of course nice for translators if there are only minor changes
before a release but it should not affect the quality of the English
document, especially since translators may find many minor issues during
translation (and during the freeze).
> > What reason could exist for this?
> > * Releasing a version which is not fully translated and is XML-based
> > would contain old information, I really understand this.
> The current missing file for some XML-based translations makes that
> translation unbuildable.
Ah, I didn't know this. If this is the case I completely agree with you
not to publish the file if major updates are required.
On the other side other people could easily fix this by copying English parts
into the translation to define missing labels, ..., of course without
updating translation check markers.
I for example would do this but I strongly remember your decision that
*nobody* (except you and Christian) should touch other translations for
any reason! This includes also FTBFS issues.
(Remember that also few translators just translate and do not try to
build the documents, so they do not recognize build errors.)
> There have been some important changes in partitioning and preseeding
> since the last release. I _do_ feel that doing an official release where
> translations contain outdated information is a disservice to users.
I more or less agree but also think that this needs to decided
individually for each case. If a translation doesn't update after a
minor fix in the English manual it's not very important. And it only
affects XML, not PO files!
> One of the main reasons to be a bit more strict for intermediate releases
> is to avoid a huge pile up of translation updates for the final release,
> so IMO this policy increases the chance that translations will be
> up-to-date at the time of the final release.
No, the opposite is true! As I wrote I would ensure that my translation
is never at 100% just to be solidly with translators who don't have the
possibilty to update the translation in time.
If you want a good translation status, kindly ask translators, do not
> > * Releasing a PO file based translation with a single English sentence
> > in it is not the right thing for Debian users. Instead they should
> > refer to the English version (because they maybe have no way to
> > obtain the inofficial translation from Alioth). Who cares whether
> > they are able to understand English?
> I would hope that the translators do...
This was ironical. But what about people without network access? They
require a translation in the installation-guide-i386 package from
CD/DVD. This version can also easily be indexed by Debian's
documentation systems, ...
I really, really do not understand your statement. A partly translated
document (PO based) is much much better than no document! Think about this
again, ask others, ...!!!!!
You should not penalize thousands of users just because a single
translator (or maybe two) were unable to finish it! Once you drop such a
document you likely also lost the translator.
> > * Most probably the current document is nearly impossible to
> > understand if there are one or two untranslated strings in it. So I
> > suggest to drop the English version as well. Not everyone is able to
> > process in memory/brain such a huge document, which one only
> > understood if it completely parsed.
> We all know that the manual can be improved. If someone does not
> understand what a part of a manual is about, they are free to ask (as
> several people have done in the past; and they've even received answers).
This was also ironical and referred to the fact, that you wanted to
exclude translation with only minor missing updates.
> A translator is even free to not translate that part. If someone does
> decide to do that, I would suggest to discuss that with me so I can take
> it into account when writing these updates.
This seems inconsistent which what you wrote above, where you wanted to
drop a translation if it's not (nearly) up-to-date.
> For PO based translations the correct way to "not translate" a particular
> part of the manual is to copy the msgid to the msgstr. If you don't then
> I cannot distinguish between "outdated" and "not translated on purpose".
PO files are never outdated, because English text will be used for fuzzy
messages! Using English mgstr would make it more difficult if a other
translator wants later to translate all remaining parts.
Why do you think untranslated messages on purpose are better as
a "lack of time" reason? Or why do you not let translators decide
whether a document should be released or net (defaulting to yes, in case
the translator is MIA).
> > * We don't need any feedback for translations. Let's hide these so that
> > nobody (except the original translator) is able to improve it.
> If you want feedback, then you can get it based on the Alioth website as
> well as the released packages.
This is not an option for many users! It's also not (locally) indexed
(only per Google, ...). Why not drop all installation-guide* packages,
users could (following your proposal) always use the Alioth version!
Do you really see no conflict in your statements?
(On the other side most users don't care at all. The German translation
of the Debian website gets not more than one or two remarks per year.)
> I feel that users have a right to expect a released translation to contain
> the same information as the original and I feel that is more important
> than this point.
A simple message similar to "This document is not fully translated" at
the beginning of the document would inform the user as well. The user
could open the English document in this case. It also is no argument at
all for PO file based translations.
Did you just forgot to mention that you want to drop outdated XML based
> > * There is no need to test the disk requirements of all translations!??
> No, there is not.
OK. Most probably there are too many last minute changes in other
packages to ensure that a fixed set of packages can be found on the
first (or second, ...) CD.
> > Mhm, German is missing again ...
> That is because German is fully up to date. I do realize now that there is
> a discrepancy between the way I listed the XML-based translations and the
> PO-based translations. Sorry for that.
> > Please consider also the case the you're not able to complete the
> > manual before the final release because of personal reasons such as
> > illness.
> That is cheap.
> If there are personal reasons why a translation cannot be completed, the
> translator is of course free to reply to my "call for update
> translations" and discuss the situation.
Why do you care about the reasons?