Julian Foad wrote on Wed, 07 Mar 2018 20:42 +0000:
> Daniel Shahaf wrote:
> > Mattias Engdegård wrote on Wed, 07 Mar 2018 16:35 +0100:
> >> The gettext manual recommends:
> >> Translatable strings should be limited to one paragraph; [...]
> > There is no reason the translator should have to re-read the entire message.
> > [...]
> > 3. Display a wdiff3 of (old = rN, theirs = rM, mine = the po file).
> > .... as I described last year:
> > https://mail-archives.apache.org/mod_mbox/subversion-dev/201702.mbox/%3C20170205204820.GA6737%40fujitsu.shahaf.local2%3E
> > I think the above workflow makes a lot more sense than paragraph-wise
> > translations.
> > [...]
> > P.S. I'm genuinely curious to know why the gettext manual doesn't
> > recommend this approach already. Version control predates
> > gettext, doesn't it?
> When a translator tells me what we need to do to "make life for
> translators somewhat bearable" I am inclined to prioritize that. It is
> simple to do and costs us little and apparently is important in real
> life. In other words it is important for the Subversion project.
> My technical sympathies are with the arguments about how a better
> solution can be built on version control and word-wise diff tools, but I
> do not want us to refuse to comply with the simple solution while we are
> waiting for the better solution to be in widespread use.
To be clear, I described two solutions:
2. Having the translator run plain old 'svn diff -r N:M' and eyeball the output.
I agree that #1 is a blue sky idea --- more precisely, one with a non-
trivial bootstrap cost --- and as such it is reasonable to reject it,
following "don't let the best be the enemy of the good".
#2, however, doesn't have a high bootstrap cost, and I am not sure whether
you considered it or glossed over it. If you've considered it and still
prefer the paragraphisation approach, that's fine; I just wanted to make
sure it hadn't gotten overlooked. The decision lies, as always, with
the people who do the work.
> Therefore I plan to go ahead, if there are no strong objections.