[SciPy-dev] Some Feedback

Fernando Perez Fernando.Perez at colorado.edu
Sun Nov 13 14:17:18 CST 2005


Jonathan Taylor wrote:

>>- trac by default ties logins to the svn developers with commit access.  We
>>want non-developers with logins.  Robert suggested a plugin which may take
>>care of this issue.
> 
> 
> I think you are mistaken.  By default your userbase is just a htpasswd
> file that you can easily add anyone too without giving them subversion
> access.

I am certainly no trac expert, and so far I'd just been using it as it was 
setup in the virtualmin environment that scipy offers, so feel free to correct 
me.  But from poking around with the ipython 'virtual root' account in the 
virtualmin host, I don't see a way to create new users for trac who are NOT 
svn users.  This may be a restriction imposed by the virtualmin environment, 
and there may be a workaround.  If you know of one, by all means let me know, 
as this would make it easier to take care of this issue without the 
multiplication of wikis which you (understandably) are worried about.

Even if the webmin interface that the project leads have access to doesn't 
offer this, we could always do it at the command line.  But even via that 
channel I haven't been able to find a mechanism to create, say users called

ipwiki.user1
ipwiki.user2

etc, who would NOT be part of the svn user list.

I tried making such a user with the webmin interface but WITHOUT SVN access, 
and even after adding the user to the trac-admin permissions, it still can't 
login as that user into the Trac wiki.

In summary: not being by a long shot a trac expert, I'm pretty lost here, in 
particular because I don't have full apache access on this box, only what the 
virtualmin environment gives me.  I'm trying to understand this not only for 
ipython, but because the access I have is the same Travis has for the scipy 
virtual host.

If you know the magic incantation to set up all of this reasonably (and in 
particular, without requiring real root access so that we don't become a pest 
on Joe Cooper's back), I'm willing to continue trying until we sort this out.

Best,

f




More information about the Scipy-dev mailing list