Re: [Boost-testing] National language support in svn breaks regression.py

From: Beman Dawes (bdawes_at_[hidden])
Date: 2007-09-23 17:51:08

Nicola Musatti wrote:
> Vladimir Prus wrote:
>> Nicola Musatti wrote:
>>
>>> Hallo,
>>> I just tried running regression.py, but after executing all the tests it
>>> aborts with an error message:
>>>
>>> # Executing SVN command "svn --non-interactive --username=user info
>>> D:\boost\svn
>>> \trunk\boost >D:\boost\svn\trunk\svn_info.txt"
>>> Traceback (most recent call last):
>>>    File "D:\boost\svn\trunk\regression.py", line 1037, in <module>
>>>      commands[ command ]( **accept_args( args ) )
>>>    File "D:\boost\svn\trunk\regression.py", line 831, in regression
>>>      collect_logs( tag, runner, platform, user, comment, incremental,
>>> dart_server
>>> , proxy, [] )
>>>    File "D:\boost\svn\trunk\regression.py", line 664, in collect_logs
>>>      revision = re.search( 'Revision: ([0-9]*)', svn_info ).group(1)
>>> AttributeError: 'NoneType' object has no attribute 'group'
>>>
>>> This is due to the fact that I'm running the italian version of Windows
>>> XP and on my computer Subversion defaults to italian messages, which
>>> breaks the regexp which collects the svn revision.
>> Ehm, regexp? I think that the right way to get version number of SVN
>> checkout is this:
>>
>>         $ svnversion
>>
>> that outputs a plain number. Is something fancier used by regression.py?
> 
> The output of svn info is redirected to a file and the revision number 
> is retrieved with a regexp. If the revision number is all it's ever 
> wanted, your solution looks a little more solid.

svnversion fails if there are Unicode names in the tree, such as are 
sometimes left over from filesystem tests. But that shouldn't really 
happen, so I'll change the scripts to give svnversion a try to see if it 
solves your problem.

It may be tomorrow morning before I get a chance to update the scripts.

Thanks,

--Beman