CG 5.8.3: updating committish of lsr snippets

classic Classic list List threaded Threaded
13 messages Options
Reply | Threaded
Open this post in threaded view
|

CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
Two months ago I discussed with Francisco a problem with committish and
LSR snippets:
http://lilypond-translations.3384276.n2.nabble.com/Doc-it-revision-of-snippets-in-LM-please-backport-to-2-14-td6600646.html

(read last messages in particular)


Information in CG 5.8.3, Updating translation committishes, is Ok for
people who have push access.
For those who haven't, a different method should be recommended.
Otherwise the maintenance scripts will complain about bad objects (non
existing committishes).

In the thread above Francisco proposed a solution.
Please think about what's the best solution and update CG.

Thanks,
Federico


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
Il 02/10/2011 12:10, Federico Bruni ha scritto:

> Two months ago I discussed with Francisco a problem with committish
> and LSR snippets:
> http://lilypond-translations.3384276.n2.nabble.com/Doc-it-revision-of-snippets-in-LM-please-backport-to-2-14-td6600646.html 
>
>
> (read last messages in particular)
>
>
> Information in CG 5.8.3, Updating translation committishes, is Ok for
> people who have push access.
> For those who haven't, a different method should be recommended.
> Otherwise the maintenance scripts will complain about bad objects (non
> existing committishes).
>
> In the thread above Francisco proposed a solution.
> Please think about what's the best solution and update CG.
>
> Thanks,
> Federico

Ok, nobody is interested in this issue.

Could you please tell me which committish should I use to avoid the bad
object warning?  Also, if I don't fix the committish, I won't be able to
update the translation in the future...

I've tried:

cd Documentation/it/texidocs
sed -i
's/514674cb00c18629242dfcde0c1a4976758adc56/34607d3e36a93030690ccd780a7ffce621ca1e0f/g'
*.texidoc

But it doesn't seem to be the right committish.
Please help me, I know nothing about Git.


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Francisco Vila
2011/10/9 Federico Bruni <[hidden email]>:
> I've tried:
>
> cd Documentation/it/texidocs
> sed -i
> 's/514674cb00c18629242dfcde0c1a4976758adc56/34607d3e36a93030690ccd780a7ffce621ca1e0f/g'
> *.texidoc
>
> But it doesn't seem to be the right committish.

What exactly did you do to find out this committish? That could be
right, it matches a commit by you in which you run makelsr.py, so it
shouldn't give any "bad object" messages.

--
Francisco Vila. Badajoz (Spain)
www.paconet.org , www.csmbadajoz.com


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
Il 10/10/2011 00:28, Francisco Vila ha scritto:

> 2011/10/9 Federico Bruni<[hidden email]>:
>> I've tried:
>>
>> cd Documentation/it/texidocs
>> sed -i
>> 's/514674cb00c18629242dfcde0c1a4976758adc56/34607d3e36a93030690ccd780a7ffce621ca1e0f/g'
>> *.texidoc
>>
>> But it doesn't seem to be the right committish.
> What exactly did you do to find out this committish? That could be
> right, it matches a commit by you in which you run makelsr.py, so it
> shouldn't give any "bad object" messages.
>
I checked the log of files in snippets/ that I've translated and I took
the commit of the last run of makelsr.py

I haven't any bad object message.

But if I run:
make ISOLANG=it check-translation

I get a strange diff (see file attached).
I think there's something wrong...


it_check-translation.diff (42K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Francisco Vila
2011/10/10 Federico Bruni <[hidden email]>:

>>> But it doesn't seem to be the right committish.
>>
>> What exactly did you do to find out this committish? That could be
>> right, it matches a commit by you in which you run makelsr.py, so it
>> shouldn't give any "bad object" messages.
>>
> I checked the log of files in snippets/ that I've translated and I took the
> commit of the last run of makelsr.py
>
> I haven't any bad object message.
>
> But if I run:
> make ISOLANG=it check-translation
>
> I get a strange diff (see file attached).

This output suggests that a further run of makelsr.py has been made
after the commitish you took.  When I watch the version history, I see
several commits labelled "run makelsr" after 34607d3e36a.  Every
change to snippets that are more recent than  34607d3e36a will show in
your check-translation output.  Do not use any committish that is
older than just one single commit: the one in which you committed the
"run makelsr" results.

In other words:
1.- update your texidoc files, run makelsr, committ these results,
write down this commit ID
2.- use this commit ID to update your texidoc files again, this time
change the committish only. Commit these changes.

This gives a pair of commits, (A) one for makelsr and (B) one to
update committish in texidocs.  The committish you put in step (B) is
the commit ID of (A).

I don't see a reason why this shouldn't work as patches-you-send as
well.  Test again and be patient! I could be wrong.
--
Francisco Vila. Badajoz (Spain)
www.paconet.org , www.csmbadajoz.com


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
I've almost gave up here.
I don't have skills nor time to spend on this issue, sorry.

The attached patch fixes the bad object error.
I have no idea if it's correct and how the updating of .texidoc files works.
make check-translation still looks weird to me, but I won't care for now....



What I've done
=========

Which is the committish that I should have used in my .texidoc files
when I committed them?
The answer should be: the commit id (my commit) of the makelsr that
processed them. Right?

So, I checked the log of each file in Documentation/snippets.
I noticed that most of them had this committish (c3b51...), so I
substitued the bad object:

sed -i
's/514674cb00c18629242dfcde0c1a4976758adc56/c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c/g'
*.texidoc

Then I fixed the few files that are related to a different committish.
This is the final committish situation in Documentation/it/texidocs:

$ cat * | grep committish
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: 34607d3e36a93030690ccd780a7ffce621ca1e0f
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: 34607d3e36a93030690ccd780a7ffce621ca1e0f
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c
%% Translation of GIT committish: 34607d3e36a93030690ccd780a7ffce621ca1e0f


0001-Doc-it-fix-the-wrong-committishes-bad-object-in-.tex.patch (15K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Francisco Vila
2011/10/23 Federico Bruni <[hidden email]>:
> I've almost gave up here.
> I don't have skills nor time to spend on this issue, sorry.

I understand it.  It is probably my fault, to be unable of explaining
it more clearly. _I_ am sorry.

> The attached patch fixes the bad object error.
> I have no idea if it's correct and how the updating of .texidoc files works.
> make check-translation still looks weird to me, but I won't care for now....
>
>
>
> What I've done
> =========
>
> Which is the committish that I should have used in my .texidoc files when I
> committed them?
> The answer should be: the commit id (my commit) of the makelsr that
> processed them. Right?

Right.  Use your commit, don't look for another old commit.

> So, I checked the log of each file in Documentation/snippets.

Mmm... the commit that appears in the log of a texidoc file is NOT the
commit of the latest makelsr run which was made from that texidoc
udpate.  You edit the texidoc, then run makelsr, this will make
changes to Documentation/snippets with your new strings, commit this
and that is the commit you should put in your texidocs.

> I noticed that most of them had this committish (c3b51...), so I substitued
> the bad object:
>
> sed -i
> 's/514674cb00c18629242dfcde0c1a4976758adc56/c3b519f0dd5ff0f8ccfc9a39ed1fe8df8b43741c/g'
> *.texidoc
>
> Then I fixed the few files that are related to a different committish.
> This is the final committish situation in Documentation/it/texidocs:

I will apply it [when at home] and let's see.

I think that as long as the docs compile, your translations are there
and up to date, and we have not too many warnings, it is enough
headache for a translator what you have done.  Remember that a
so-called fatal error caused by a bad object is not really fatal. It
only means that you will not be able to track this file's recent
history in an easy way.

If you eventually reach an acceptable status in your texidoc
committishes, you'll find that doing a frequent "makelsr + commit +
retouch texidoc committishes" sequence is something you can get used
to and worries will become fewer and smaller with time.

Thanks for your time.
--
Francisco Vila. Badajoz (Spain)
www.paconet.org , www.csmbadajoz.com


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
Hi Francisco,

Il 24/10/2011 10:17, Francisco Vila ha scritto:
> I will apply it [when at home] and let's see.
>

you haven't applied it yet, right?


> I think that as long as the docs compile, your translations are there
> and up to date, and we have not too many warnings, it is enough
> headache for a translator what you have done.  Remember that a
> so-called fatal error caused by a bad object is not really fatal. It
> only means that you will not be able to track this file's recent
> history in an easy way.

Yes, I know and I'm a bit worried about this scenario.

I hoped that finding the "right commit" was an easy job, at least for an
experienced contributor. I hoped that it could be fixed, so I could go
on with new translations (I'd like to start NR).


> If you eventually reach an acceptable status in your texidoc
> committishes, you'll find that doing a frequent "makelsr + commit +
> retouch texidoc committishes" sequence is something you can get used
> to and worries will become fewer and smaller with time.

Yes, sure, I'll learn by experience.
But please let's try to fix it now.

Thanks,
Federico


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Francisco Vila
2011/10/31 Federico Bruni <[hidden email]>:
> Hi Francisco,
>
> Il 24/10/2011 10:17, Francisco Vila ha scritto:
>>
>> I will apply it [when at home] and let's see.
>>
>
> you haven't applied it yet, right?

Now it is; please pull, test it and tell me.

--
Francisco Vila. Badajoz (Spain)
www.paconet.org , www.csmbadajoz.com


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
Il 31/10/2011 21:05, Francisco Vila ha scritto:
>> you haven't applied it yet, right?
> Now it is; please pull, test it and tell me.

Ok, thanks.
Well, I had tested it before and I knew what it would have happened.

'make check-translation' is a mess now.

If you run this command you'll see what I mean:

cd Documentation
make ISOLANG=it NO_COLOR=1 check-translation | less


Is it normal?
I guess it's not. There's a lot of garbage.
I've tried the check-translation of spanish and in most (all?) of cases
the only change for each file in snippets/ is the committish.


BTW, it would be much better if check-translation could check only the
modifications in the texidoc="" and doctitle="" parts of the snippets
(the parts that must be translated).
Is it possible? (question for Python developers)


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Francisco Vila
2011/11/1 Federico Bruni <[hidden email]>:
> 'make check-translation' is a mess now.
>
> If you run this command you'll see what I mean:
>
> cd Documentation
> make ISOLANG=it NO_COLOR=1 check-translation | less

I usually leave colors on and use less -R instead.

> Is it normal?

I do not see this output especially abnormal.  As you know, from this
output you will obtain one of

 -- changes to lilypond code.  You should ignore it.
 -- changes to translations doctitleXX or texidocXX other than yours.
Ignore those, too.
 -- changes to header strings in English other than texidoc or
doctitle. Ignore them.
 -- changes to GIT committish strings either yours or others'.  They
reflect makelsr runs made by others.  I think in general you can
ignore them.
 -- changes that show deleted snippets. Delete yours.

That leaves us with the only important changes: those in texidoc and
doctitle English strings.  Update your translations accordingly, run
makelsr, commit, use this commit ID to update the committishes in your
translations, commit this. Done.

BTW we don't have a clear warning of the existence of new
translation-needing snippets.  I usually see English paragraphs in
Spanish docs and that is a sign, but it is not pleasing to find them
in already released docs.

> I guess it's not. There's a lot of garbage.
> I've tried the check-translation of spanish and in most (all?) of cases the
> only change for each file in snippets/ is the committish.

Would it be OK if I managed to find a committish that gives a terse
check-translation output for Italian?  I think if I found one for
Spanish, that is certainly possible.  Be aware, you'll lose any
tracking information.

> BTW, it would be much better if check-translation could check only the
> modifications in the texidoc="" and doctitle="" parts of the snippets (the
> parts that must be translated).

I agree.

> Is it possible? (question for Python developers)

the output of check-translation is made by diff, and diff is blind to
syntax structures. It is certainly feasible because we are already
parsing our translations when we run makelsr so that it puts them into
snippets.  This is far beyond my skills or my available time. Or both.

--
Francisco Vila. Badajoz (Spain)
www.paconet.org , www.csmbadajoz.com


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Federico Bruni-2
Il 01/11/2011 20:32, Francisco Vila ha scritto:

> I do not see this output especially abnormal.  As you know, from this
> output you will obtain one of
>
>   -- changes to lilypond code.  You should ignore it.
>   -- changes to translations doctitleXX or texidocXX other than yours.
> Ignore those, too.
>   -- changes to header strings in English other than texidoc or
> doctitle. Ignore them.
>   -- changes to GIT committish strings either yours or others'.  They
> reflect makelsr runs made by others.  I think in general you can
> ignore them.
>   -- changes that show deleted snippets. Delete yours.
>
> That leaves us with the only important changes: those in texidoc and
> doctitle English strings.  Update your translations accordingly, run
> makelsr, commit, use this commit ID to update the committishes in your
> translations, commit this. Done.
>

This means that I'll have to skim all the garbage and be careful not to
miss the updates that I should not ignore.
It's feasible, but not handy.


> BTW we don't have a clear warning of the existence of new
> translation-needing snippets.  I usually see English paragraphs in
> Spanish docs and that is a sign, but it is not pleasing to find them
> in already released docs.
>

Ok, noted.


>> I guess it's not. There's a lot of garbage.
>> I've tried the check-translation of spanish and in most (all?) of cases the
>> only change for each file in snippets/ is the committish.
>
> Would it be OK if I managed to find a committish that gives a terse
> check-translation output for Italian?  I think if I found one for
> Spanish, that is certainly possible.  Be aware, you'll lose any
> tracking information.
>

I don't want to lose tracking information: it's more important than
having to deal with the garbage output of check-translation.

On the other hand, I thought (and still think) that the "right
committish", which keeps tracking information AND gives a terse
check-translation output, should exist for these files.
If we can't find it, I hope that I'll fix the problem when I'll have the
chance to update the files.


Reply | Threaded
Open this post in threaded view
|

Re: CG 5.8.3: updating committish of lsr snippets

Francisco Vila
2011/11/3 Federico Bruni <[hidden email]>:
> On the other hand, I thought (and still think) that the "right committish",
> which keeps tracking information AND gives a terse check-translation output,
> should exist for these files.

Not necesarily, because snippets change. That is where garbage comes
from.  You see garbage once, and then if you update the committishes
you should not see that garbage again. You'll see another, different
garbage.

We need a terse output that shows only changes in texidocs and titles.

--
Francisco Vila. Badajoz (Spain)
www.paconet.org , www.csmbadajoz.com