msguniq in po-update.sh

[Mattias Engdegård <ma...@acm.org>]

Sorry about messing up the build with my ill-considered merge of r1841318 (sv.po). Instead of attempting to merge from 1.11.x, which is what caused the duplicated msgids, I should just have copied the entire file from that branch and be done with it. That was just careless on my part.

However, r1841716 and r1841717 (adding msguniq to po-update.sh and running it) may have been made in haste. The latter reformatted the entire file, making it impossible to see what had actually changed; I had to write a script to normalise the strings in order to make sense of that change, and the line history is completely gone.

I'll replace sv.po in trunk with that in 1.11.x, which does not contain any duplicated msgids. Would you agree to revert r1841716, on the grounds that having msguniq as part of the script is somewhat invasive and dangerous? Not only does it reformat the file, but I'm far from sure what it does with duplicated msgids, and whether that is the right course of action in all cases. Surely human intervention is safer.

By the way, my thanks to Julian Foad for mending another of my mistakes (missing newlines). These errors are usually linted out prior to committing; I have no excuse for not doing that.

Re: msguniq in po-update.sh

[Julian Foad <ju...@apache.org>]

Stefan Sperling wrote:
> Mattias Engdegård wrote:
>> However, r1841716 and r1841717 (adding msguniq to po-update.sh and running
>> it) may have been made in haste. [...]
>> 
>> I'll replace sv.po in trunk with that in 1.11.x, which does not contain any
>> duplicated msgids. Would you agree to revert r1841716, [...]
> 
> Sure. [...] I am fine with you reverting my
> changes and providing a better solution. (I won't have time to
> revert those changes myself over the coming days, sorry.)

I will make those changes right now, and remove the backport proposal.

>> By the way, my thanks to Julian Foad for mending another of my
>> mistakes (missing newlines). These errors are usually linted out prior
>> to committing; I have no excuse for not doing that.

You're welcome. It's easily done.

-- 
- Julian

Re: msguniq in po-update.sh

[Stefan Sperling <st...@elego.de>]

On Sun, Sep 23, 2018 at 11:06:40PM +0200, Mattias Engdegård wrote:
> Sorry about messing up the build with my ill-considered merge of r1841318
> (sv.po). Instead of attempting to merge from 1.11.x, which is what caused the
> duplicated msgids, I should just have copied the entire file from that branch
> and be done with it. That was just careless on my part.
> 
> However, r1841716 and r1841717 (adding msguniq to po-update.sh and running
> it) may have been made in haste. The latter reformatted the entire file,
> making it impossible to see what had actually changed; I had to write a
> script to normalise the strings in order to make sense of that change, and
> the line history is completely gone.
> 
> I'll replace sv.po in trunk with that in 1.11.x, which does not contain any
> duplicated msgids. Would you agree to revert r1841716, on the grounds that
> having msguniq as part of the script is somewhat invasive and dangerous? Not
> only does it reformat the file, but I'm far from sure what it does with
> duplicated msgids, and whether that is the right course of action in all
> cases. Surely human intervention is safer.

Sure. Whichever solution works for you is fine for me.

I just committed what seemed like a working solution to allow me to
make progress beyond a build failure. The use of msguniq was suggested
in a source code comment in gettext close to the crash I ran into,
so I tried it and it resolved the issue at hand for me.

But I am not involved in translations and am unaware of the finer
points that need consideration. I am fine with you reverting my
changes and providing a better solution. (I won't have time to
revert those changes myself over the coming days, sorry.)