[AG-TECH] How to build from the CVS

[Douglas Kosovic]

Hi Socrates,

> I was wondering what the preferred way of building the AGTk v2.3 is:
>
> a) I have been using the AccessGrid/tools/CvsSetup.py script (as described
> in the website) after a CVS checkout. With that the toolkit starts and 
> runs
> fine.
> b) Our colleagues at Southampton have been using the
> AccessGrid/packaging/BuildSnapshot.py script. This method also results in 
> an
> operational toolkit.
>
> So, could someone shed some light on whether (a) or (b) is most preferred
> and any possible side-effects of using one or the other method?
>
> We're both using Linux: RedHat 9.0 for case (a), Fedora Core 2 for (b).
>
> Any comments appreciated.

In packaging/linux/fedora/AccessGrid.spec.in I put a python >= 2.2.3 
requirement which is the only thing stopping option (b) from building on 
Redhat 9.0. RedHat includes python 2.2.3 with RHEW3 and FC1, while RH9 comes 
with python 2.2.2 which has the broken XML parser.

For the moment to build the RedHat/Fedora AGTk 2.3 RPMs, you would have to 
use the src tarball from http://www.itee.uq.edu.au/~agn/ag_fedora/ as the 
changes haven't been checked into CVS yet. Build instructions can be found 
on that page.

As to using either option (a) or (b), it's a matter of preference and 
whichever is more suitable for the situation. Probably the only difference 
is that the RPM comes with optimized python bytecode files (*.pyo) and 
Gnome/XDG menu entries.


Cheers,
Doug