Mailing List Archive

2010/10/1 ikke <ikke at iki.fi>:
> I suggest someone would do a separate commit only for removing the
> code line reference  locations in *.ts files. They just generate lot
> of distracting noise while doing svn diff after doing lupdate.

Yes, they do generate a lot of noise, but they're needed for Qt
linguist to show the related source code. In many cases you would not
be able to do a correct/good translation of a string, without looking
at the context of the string. The same word can have a different
meaning in two different contexts.

But why do you consider this a problem? If it's because you want to
review your own patch, then just filter out the unwanted lines with
grep or similar, when you're doing the review of your patch.

> If that's not ok for all, please do it at least for the finnish
> translations now that there  is patch coming in (9039).

I'm sorry, but I don't think that it would be a good idea...even if we
(or you) generate the Finnish *.ts files with the "-locations none"
argument, then once someone else (me, knightr or another Finnish
contributor) updates the .ts file, the code line references will get
added again...and knightr and I will almost certainly begin to run
lupdate regularly on all languages to keep the files somewhat up to
date.

For now, the policy is that we run with the "-locations absolute"
argument, since this is considered a helping hand for translators.
Also, the size of the patch shouldn't really have any negative impact
on the translators work (unless of course if the translator is
reviewing the patch in detail himself - but this shouldn't be needed,
knightr and I will do this if needed, before committing to SVN).

Best Regards
Kenni

On Fri, Oct 1, 2010 at 11:25 PM, Kenni Lund <kenni at kelu.dk> wrote:
> 2010/10/1 ikke <ikke at iki.fi>:
>> I suggest someone would do a separate commit only for removing the
>> code line reference  locations in *.ts files. They just generate lot
>> of distracting noise while doing svn diff after doing lupdate.
>
> Yes, they do generate a lot of noise, but they're needed for Qt
> linguist to show the related source code. In many cases you would not
> be able to do a correct/good translation of a string, without looking
> at the context of the string. The same word can have a different
> meaning in two different contexts.

I agree they are useful, I use the feature also in linguist. But
that's easily overcome by having the instructions in the translation
wiki to do at first:

lupdate -locations relative

each time one starts the translation. After editing, it's the cleanest
way to review the changes by doing:

lupdate -locations none
svn diff | less

then sending the patch, and now the patch is the most minimal in size,
and easy to see what actually changed.

> But why do you consider this a problem? If it's because you want to
> review your own patch, then just filter out the unwanted lines with
> grep or similar, when you're doing the review of your patch.

This is not matter of life and death to me :) , but at least it feels
the least effort way of working for me. It's about review, and
tracking the code changes is easiest when the extra avoidable noise is
in minimum.

>> If that's not ok for all, please do it at least for the finnish
>> translations now that there  is patch coming in (9039).
>
> I'm sorry, but I don't think that it would be a good idea...even if we
> (or you) generate the Finnish *.ts files with the "-locations none"
> argument, then once someone else (me, knightr or another Finnish
> contributor) updates the .ts file, the code line references will get
> added again...and knightr and I will almost certainly begin to run
> lupdate regularly on all languages to keep the files somewhat up to
> date.
>
> For now, the policy is that we run with the "-locations absolute"
> argument, since this is considered a helping hand for translators.

Matter of the instructions in the wiki really.

> Also, the size of the patch shouldn't really have any negative impact
> on the translators work (unless of course if the translator is
> reviewing the patch in detail himself - but this shouldn't be needed,

Doesn't matter as long there is one person doing the translations,
but if people would like to cross review others work it starts making
difference. Smaller the patch, easier it is to review.

> knightr and I will do this if needed, before committing to SVN).

Ok, this issue is much of matter of taste, and this was just a
suggestion. I can work either way, thanks for applying the patch
anyway!

> Best Regards
> Kenni