Re: [wx-translators] Using Rosetta to handle translations
Vadim Zeitlin
JM> I think using rosetta might give the wxWidgets templates a wider
JM> visibility and, in the end, more submitted translations. Rosetta is
JM> actively being used to translate Ubuntu. One of the key features is that
JM> it hides the PO format behind a clean web-based UI so that people can
JM> translate stuff without knowing how the PO format or a PO file editor
JM> works. It also has an import/export mechanism for the more advanced
JM> translators.
This is actually the only question I have about it: how are we going to
handle conflicts between the changes done to .po files in our cvs and the
changes done using Rosetta? As you surely know, merging .po files using the
usual tools is all but impossible as the changes in the lines numbers
confuse them hopelessly. So what are we going to do if one of the existing
translators sends me an updated version of the .po file (which is how it
works so far) but it was already modified using Rosetta?
But otherwise, I'd be really glad to use Rosetta for wx translations as it
definitely should result in more people helping with them.
Thanks!
VZ
Jordi Mallach
Hello list,
I'm in charge of handling the translation upload requests in Rosetta,
Ubuntu's translation portal <https://launchpad.net/rosetta/>.
This week, someone requested to upload the wxstd.pot file to our system,
and as our policy says that we should only upload templates when the
upstream authors are informed and agree, I went to talk to the wxwidgets
developers on the IRC channel.
I found that the upload had not been requested or discussed by the
developers, but they were open to the idea of using Rosetta in any
case, as it might bring some benefits to both the developers in charge
of dealing with wxWidgets translations, and the translators themselves.
I think using rosetta might give the wxWidgets templates a wider
visibility and, in the end, more submitted translations. Rosetta is
actively being used to translate Ubuntu. One of the key features is that
it hides the PO format behind a clean web-based UI so that people can
translate stuff without knowing how the PO format or a PO file editor
works. It also has an import/export mechanism for the more advanced
translators.
The maintainers can also benefit by fetching all the translations as
they are with a single click, instead of collecting them by email, etc.
Mart Raudsepp has asked me to propose Rosetta on this list, so here we
are. Feel free to ask any questions about Rosetta, I'll try my best to
answer them all. You can learn more about it in the FAQ:
https://wiki.ubuntu.com/RosettaFAQ
Thanks,
Jordi
--
Jordi Mallach Pérez -- Debian developer http://www.debian.org/
jo...@sindominio.net jo...@debian.org http://www.sindominio.net/
GnuPG public key information available at http://oskuro.net/
ABX
> I think that if we decide to use Rosetta, we should indeed make it the
>*only* way to modify wx translations, otherwise we're bound to have
>problems. So the question is what do the people -- translators -- think
>about it. Please post your opinions to wx-translators!
I took a look into interface at https://launchpad.net/products/wxwidgets/2.6/
and after logging in I believe I find poEdit more comfortable for the simple
reason that it opens source code with context of the phrase which is very
usefull for some of the items. I see some benefits in Rosetta interface too
(for team work on single translation) but browsing the source code is not
handy. So I'm against choosing Rosetta as the only system but as alternate
input it would be valuable.
ABX
Vadim Zeitlin
A> Vadim Zeitlin <va...@wxwindows.org>:
A> > I think that if we decide to use Rosetta, we should indeed make it the
A> >*only* way to modify wx translations, otherwise we're bound to have
A> >problems. So the question is what do the people -- translators -- think
A> >about it. Please post your opinions to wx-translators!
A>
A> I took a look into interface at https://launchpad.net/products/wxwidgets/2.6/
A> and after logging in I believe I find poEdit more comfortable for the simple
A> reason that it opens source code with context of the phrase which is very
A> usefull for some of the items. I see some benefits in Rosetta interface too
A> (for team work on single translation) but browsing the source code is not
A> handy. So I'm against choosing Rosetta as the only system but as alternate
A> input it would be valuable.
Well, there have been no other opinions so I'm not sure about how should
we proceed. We don't have any solution to using both Rosetta and poEdit in
parallel and while we don't want to abandon the possibility to use poEdit,
we'd like to use Rosetta as well. So we're in a dead end... Maybe the best
would be to only use Rosetta for some languages (i.e. those without active
translators) and leave the translators for the other languages use whatever
tools they prefer?
Thanks,
VZ
Vadim Zeitlin
JM> On Wed, Jun 07, 2006 at 09:22:44PM +0200, Vadim Zeitlin wrote:
JM> > This is actually the only question I have about it: how are we going to
JM> > handle conflicts between the changes done to .po files in our cvs and the
JM> > changes done using Rosetta? As you surely know, merging .po files using the
JM> > usual tools is all but impossible as the changes in the lines numbers
JM> > confuse them hopelessly. So what are we going to do if one of the existing
JM> > translators sends me an updated version of the .po file (which is how it
JM> > works so far) but it was already modified using Rosetta?
JM>
JM> That's a difficult question. The short answer is: choose which you want
JM> to give more weight and use msgcat --use-first.
I've never used this msgcat option so I might be missing something but I
don't see how would this work when we don't even know where the latest
version is (in Rosetta or our cvs), and so don't know which one should be
first.
JM> Ideally, though, translations would not be sent to you directly, but
JM> uploaded to Rosetta. You would periodically fetch the tarball with all
JM> the translations and mass-commit them in SVN.
Of course, switching to Rosetta exclusively would solve the problem. But
then we should take into account that current translators don't use
Rosetta. I don't have any experience with Rosetta myself but even assuming
it's totally great, I'm sure that some of them would still prefer to work
as before, just because there is no such thing as unanimous agreement.
And while I don't want at all for the translations to be sent to me
(they're not code and I can't verify them (beyond checking "msgfmt -c -v"
output) anyhow so what's the point?). But the very practical problem we're
going to have is that someone is still going to send me a new .po file,
whether I want it or not, and I won't be able to commit it to the cvs
because it would overwrite the Rosetta changes. And I just don't see how am
I going to be able to track it. Without a tool to intelligently compare and
merge .po files it seems we're bound to have problems. I wonder how do the
other projects deal with this...
JM> This problem is not specific to Rosetta or any other web-based
JM> translation system:
No, definitely not. I just hoped that maybe Rosetta had some smart way of
solving it.
JM> if you, with some bad luck, got two independent translations of wxstd
JM> to the same language at the same time (ie, a new translation started at
JM> the same time by two different people -- I've seen this happen in some
JM> other projects), you'd have the very same problem.
Yes, indeed. Which is why I try to coordinate people (and why this mailing
list exists). But it's going to be more difficult to do if/when we have
many more translators using Rosetta.
JM> > But otherwise, I'd be really glad to use Rosetta for wx translations as it
JM> > definitely should result in more people helping with them.
JM>
JM> I'm glad to hear this. I'll stay tuned on the discussion. :)
I think that if we decide to use Rosetta, we should indeed make it the
*only* way to modify wx translations, otherwise we're bound to have
problems. So the question is what do the people -- translators -- think
about it. Please post your opinions to wx-translators!
Thanks,
VZ
Hans F. Nordhaug
Just a PS: Rosetta supports download/upload of .po-files so that you
can continue to use your favorite .po-file editor to translate - look
for example at the left menu at
https://launchpad.net/products/wxwidgets/2.6/+pots/wxstd/nb/+translate
Some other thoughts:
1) I also prefer poEdit, but because of the above I wouldn't mind[1] that
my translation (Norwegian Bokmål) was moved to Rosetta.
2) I agree wholeheartedly that *if* we select Rosetta it should be the
only way - I just hate the confusion (and double work that normally
happens) when there are multiple sources for a translation.
Sorry for the late reply - I missed this topic.
Regards,
Hans
[1] I *do* mind that Rosetta is (still) non-free, but I hope/expect
that to change.
Jordi Mallach
On Wed, Jun 07, 2006 at 09:22:44PM +0200, Vadim Zeitlin wrote:
> This is actually the only question I have about it: how are we going to
> handle conflicts between the changes done to .po files in our cvs and the
> changes done using Rosetta? As you surely know, merging .po files using the
> usual tools is all but impossible as the changes in the lines numbers
> confuse them hopelessly. So what are we going to do if one of the existing
> translators sends me an updated version of the .po file (which is how it
> works so far) but it was already modified using Rosetta?
That's a difficult question. The short answer is: choose which you want
to give more weight and use msgcat --use-first.
Ideally, though, translations would not be sent to you directly, but
uploaded to Rosetta. You would periodically fetch the tarball with all
the translations and mass-commit them in SVN.
From your comments I'm gettign the idea that in general only you have
the rights to commit translations in SVN, ie, translators don't have
direct access to the repo. If so, and if you wanted to allow people to
still be able to send the files directly to you, you'd be able to do it
by creating an empty translation team for that given language: you'd be
sure nobody would be able to translate on Rosetta, so no worries about
conflicts. I personally think that having alternative procedures is in
the end more confusing than having just one clean, easy way to do
things.
This problem is not specific to Rosetta or any other web-based
translation system: if you, with some bad luck, got two independent
translations of wxstd to the same language at the same time (ie, a new
translation started at the same time by two different people -- I've
seen this happen in some other projects), you'd have the very same
problem.
> But otherwise, I'd be really glad to use Rosetta for wx translations as it
> definitely should result in more people helping with them.
I'm glad to hear this. I'll stay tuned on the discussion. :)
Jordi
Jordi Mallach
On Thu, Jun 08, 2006 at 06:05:51AM +0200, Vadim Zeitlin wrote:
> JM> That's a difficult question. The short answer is: choose which you want
> JM> to give more weight and use msgcat --use-first.
> I've never used this msgcat option so I might be missing something but I
> don't see how would this work when we don't even know where the latest
> version is (in Rosetta or our cvs), and so don't know which one should be
> first.
msgcat will just concatenate two files, and --use-first will discard
any second version of a string that is already present in the first
file.
If you take two conflicting files and msgcat --use-fist one.po two.po,
the output will be a valid po file with all the strings included in both
files, but none repeated. So, if one.po was missing some strings which
were already done by some other translator in two.po, they will appear
in the resulting po thanks to the merge.
Another option could be using
msgmerge -C alttranslation.po translation.po template.pot
to use the secondary trnaslation just as a message catalog. If
"translation.po" was missing any string, it would use the solution in
the alternative file, if it exists.
> Of course, switching to Rosetta exclusively would solve the problem. But
> then we should take into account that current translators don't use
> Rosetta. I don't have any experience with Rosetta myself but even assuming
> it's totally great, I'm sure that some of them would still prefer to work
> as before, just because there is no such thing as unanimous agreement.
*nod*. Still, the amount of work involved in mailing a file to you and
submitting it via a webform is quite similar. You might want to enforce
it, as it is quite better from the maintainer's point of view. Everyone
using the same method helps a lot to minimise the problem described
above, too.
> And while I don't want at all for the translations to be sent to me
> (they're not code and I can't verify them (beyond checking "msgfmt -c -v"
> output) anyhow so what's the point?). But the very practical problem we're
> going to have is that someone is still going to send me a new .po file,
> whether I want it or not, and I won't be able to commit it to the cvs
> because it would overwrite the Rosetta changes. And I just don't see how am
> I going to be able to track it. Without a tool to intelligently compare and
> merge .po files it seems we're bound to have problems. I wonder how do the
> other projects deal with this...
Well, if you still got some files by email, you would be able to, as
product and template owner, upload them to Rosetta so your stuff remains
in sync.
I think the problem can only be solved with a clear translation policy
for wxwidgets and some documentation that translators can easily find in
the webpage.
> JM> This problem is not specific to Rosetta or any other web-based
> JM> translation system:
> No, definitely not. I just hoped that maybe Rosetta had some smart way of
> solving it.
Well, Rosetta tries to minimise the impact, by providing two modes of
import. When uploading a po file, you're offered to select between
"published" and "user" uploads. If published, Rosetta will prefer what
it has in the database as "accepted translation" over whatever comes in
the uploaded po file. If "user", the strngs uploaded will overwrite any
different content in Rosetta. This way you can select which set of
translations you want to give priority.
Furthermore, the "discarded" translations won't be thrown away. They
will appear as "suggestions" in the web interface, so they remain
visible in case a translator later thinks they are better solutions to
what is currently accepted.
> Yes, indeed. Which is why I try to coordinate people (and why this mailing
> list exists). But it's going to be more difficult to do if/when we have
> many more translators using Rosetta.
I feel we're trying to solve the same problem. :)
> I think that if we decide to use Rosetta, we should indeed make it the
> *only* way to modify wx translations, otherwise we're bound to have
> problems. So the question is what do the people -- translators -- think
> about it. Please post your opinions to wx-translators!
Absolutely. Please keep me in touch of how your debate goes.