Change Request: Test unicode fixes on transifex on app4

[Ricky Zhou]

Recently, translators with unicode characters in their name have been
unable to commit to some modules in Transifex.  Since these translations
need to be submitted for the F9 release and I haven't been able to
reproduce on a testing instance, I'd like to try a fix on the production
version on app4.  This should be pretty low risk (and affect nothing but
transifex for a few minutes).  The change looks something like:

command = [arg.encode('utf8') for arg in command]

after line 127 in /srv/tg/transifex/transifex/util.py.  If this doesn't
work out, I will have to look for other problems (which should once
again only affect transifex for a bit).  

Does this sound safe?  

Ricky

[Mike McGrath]

On Wed, 30 Apr 2008, Ricky Zhou wrote:

>  Recently, translators with unicode characters in their name have
> been
>  unable to commit to some modules in Transifex.  Since these translations
>  need to be submitted for the F9 release and I haven't been able to
>  reproduce on a testing instance, I'd like to try a fix on the production
>  version on app4.  This should be pretty low risk (and affect nothing but
>  transifex for a few minutes).  The change looks something like:
>
>  command = [arg.encode('utf8') for arg in command]
>
>  after line 127 in /srv/tg/transifex/transifex/util.py.  If this doesn't
>  work out, I will have to look for other problems (which should once
>  again only affect transifex for a bit).
>
>  Does this sound safe?
>  
Not doing this means some translations for some F9 content won't get
translated?  if thats the case then I'd say +1.  Seems low risk and has a
positive, direct outsomce on F9.

	-Mike

[Ignacio Vazquez-Abrams]

On Wed, 2008-04-30 at 13:07 -0400, Ricky Zhou wrote:
>  Recently, translators with unicode characters in their name have
> been
>  unable to commit to some modules in Transifex.  Since these translations
>  need to be submitted for the F9 release and I haven't been able to
>  reproduce on a testing instance, I'd like to try a fix on the production
>  version on app4.  This should be pretty low risk (and affect nothing but
>  transifex for a few minutes).  The change looks something like:
>  
>  command = [arg.encode('utf8') for arg in command]
>  
>  after line 127 in /srv/tg/transifex/transifex/util.py.  If this doesn't
>  work out, I will have to look for other problems (which should once
>  again only affect transifex for a bit).  
>  
>  Does this sound safe?   
-1

I'm not convinced that this shotgun approach is best. I recommend
applying the less-invasive attached patch and figuring out what's wrong
first.

-- 
Ignacio Vazquez-Abrams <ivazqueznet(a)gmail.com&gt;

[Ricky Zhou]

On 2008-04-30 02:12:01 PM, Ignacio Vazquez-Abrams wrote:
>  I'm not convinced that this shotgun approach is best. I
> recommend
>  applying the less-invasive attached patch and figuring out what's wrong
>  first. 
All right then, we're going to try and get a bit more information about
where the problem is coming from first (adding those debugging
statements now).  

Thanks,
Ricky

[Ignacio Vazquez-Abrams]

On Wed, 2008-04-30 at 14:18 -0400, Ricky Zhou wrote:
>  On 2008-04-30 02:12:01 PM, Ignacio Vazquez-Abrams wrote:
>  > I'm not convinced that this shotgun approach is best. I recommend
>  > applying the less-invasive attached patch and figuring out what's wrong
>  > first.
>  All right then, we're going to try and get a bit more information about
>  where the problem is coming from first (adding those debugging
>  statements now).   
This patch should fix it based on discussion in IRC.

-- 
Ignacio Vazquez-Abrams <ivazqueznet(a)gmail.com&gt;

[Ricky Zhou]

On 2008-04-30 02:31:59 PM, Ignacio Vazquez-Abrams wrote:
>  On Wed, 2008-04-30 at 14:18 -0400, Ricky Zhou wrote:
>  > On 2008-04-30 02:12:01 PM, Ignacio Vazquez-Abrams wrote:
>  > > I'm not convinced that this shotgun approach is best. I recommend
>  > > applying the less-invasive attached patch and figuring out what's wrong
>  > > first.
>  > All right then, we're going to try and get a bit more information about
>  > where the problem is coming from first (adding those debugging
>  > statements now).  
>  
>  This patch should fix it based on discussion in IRC. 
We ended up encoding all of
the arguments.  From what we concluded, this
might be a python 2.4 bug, so this was the most general way to fix it
(since other code uses run_command as well, and we want to encode as
late as possible).  

Thanks,
Ricky

[Toshio Kuratomi]

Ricky Zhou wrote:
>  Recently, translators with unicode characters in their name have
> been
>  unable to commit to some modules in Transifex.  Since these translations
>  need to be submitted for the F9 release and I haven't been able to
>  reproduce on a testing instance, I'd like to try a fix on the production
>  version on app4.  This should be pretty low risk (and affect nothing but
>  transifex for a few minutes).  The change looks something like:
>  
>  command = [arg.encode('utf8') for arg in command]
>  
>  after line 127 in /srv/tg/transifex/transifex/util.py.  If this doesn't
>  work out, I will have to look for other problems (which should once
>  again only affect transifex for a bit).  
>  
>  Does this sound safe?  
>   
Sounds safe.  But after looking at this on IRC I'm not sure that 
addresses the actual problem (it might be treating a symptom.)

Can we get a bit of logging into there so we can see what's going on?

I'd like to know what the value of the arguments being passed to 
subprocess.Popen() is and also what their types are.

-Toshio