Re: Regarding LSR translation work

[Federico Bruni]

Il 25/04/2012 12:04, Graham Percival ha scritto:
[..]
> > >  I'm trying to learn Python and I'd like to contribute to some frog
> > >  tasks requiring python (I've starred some frog issues in the tracker).
> > >  Probably this is too much difficult for a newbie.
> Actually, I've got a good suggestion on how you could get started
> with python and all this stuff.
>
> Currently, makelsr.py adds different amounts of blank lines to
> some .ly files depending on whether you run it as a full import
> vs. when you run it locally.  It would be nice if that didn't
> happen.

[I've just started looking at this.]

I think that blank lines are not the only problem.
git log shows that most of the snippets' commits are related to the 
update of committish in translated texidoc strings.
It's what we are talking about in this thread: in order to mark a 
snippet as updated, a translator updates the committish in the .texidoc 
file and run makelsr, which copies the new committish in the snippet.
So the snippet gets a new (unnecessary?) commit, which makes all the 
other translations of that snippet out-of-date.

It's a dog chasing its tail.

These committishes copied inside the snippets are needed?
IIUC, check-translation doesn't care about the committishes placed 
inside a file; it just check the latest git commit hash:

line 38 of check-translation.py:
vc_revision_parse = 'git log -1 --pretty=format:%%H %s'


What happens if we remove all the "%% Translation of GIT committish:" 
strings in the snippets and change makelsr so that it will ignore the 
first line of .texidoc files when it copies them in the snippets?

I don't believe that check-translation will stop working.
The benefits would be:

- no unuseful stuff in the commits of snippets --> cleaner git log
- cleaner git log --> less output in check-translation

Does it make sense?