Fabrice et al.,

	I think we're moving towards having eups and the entire LSST
stack use the system Python (and numpy and scipy and possibly astropy).
So your option 1 (which is what the LSST stack currently uses) will go
away.  I'm not yet sure if we will want virtualenv (your option 3); I
think most probably not.

> I see at least 3 solutions :
> 
> 1. re-building python : it's a bit awkward because eups already needs python
> and this implies python is already installed on the environnement
> before installing eups.
> Furthermore, if eups install a python version it is not compatible
> with, it will nevertheless swith to use
> it (instead of system python) and then fails.
> 
> 2. use systemp python : there's example file in HSC build script.
> The problem with this solution
> is that the install tool may have to manage conflict between system
> python libs and eups-installed python libs.
> 
> 3. use virtualenv : i think it's a good solution, but eups can't do
> this at the moment.
> 
> Maybe it would be a good idea to make eups always run with system
> python, indeed it would solve point 1. and 3.
> I can test quickly if it work by updating eups source (in peculiar
> eups executable which invoke python), and them put a ticket so that
> eups install tools take in account this improvment.
> 
> If this solution is too complicated, i will then try point 2.
> 
> Do you think this proposal is reasonable ?

-- 
Kian-Tat Lim, LSST Data Management, [log in to unmask]

########################################################################
Use REPLY-ALL to reply to list

To unsubscribe from the QSERV-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1