On Mon, May 11, 2009 at 01:36:26PM -0400, Michael Goldish wrote:
>
> ----- "Mike Burns" <mburns@xxxxxxxxxx> wrote:
>
> > Eduardo Habkost wrote:
<snip>
> > >
> > > Maybe something like (untested):
> > >
> > > for k in params.keys():
> > > os.putenv("KVM_INSTALL_%s" % (k), params[k])
> > >
> > > Are all values on 'params' guaranteed to be strings, or they can be
> > set
> > > to any python value? In the latter case, we could use
> > str(params[k]), or
> > > export only the string parameters.
> > >
> > That's a good idea. I'm not sure about whether the params are all
> > strings. I'll try it out and respin the patch after.
>
> 1. Not all params are strings -- 'depend' is a list of strings, so str(params[k]) is a good idea.
Agreed. Using str(params[k]) won't hurt.
>
> 2. Why not just pass the parameters via the command line, e.g.
> install_command = my_script.sh param1 param2
I think keeping the same convention for parameter passing for custom
scripts makes it easier to make custom scripts that behave similarly to
the predefined rules, but with just a few differences.
Also, can't the install rule be used on the cartesian configuration
file? In this case, the install parameters may be specified on a
different config rule.
For example, the Fedora project could use it like this:
cvs_server = "cvs.fedoraproject.org:/..."
variants:
- CVSF10:
cvs_branch = F10
- CVSF11:
cvs_branch = F11
- CVSRawhide:
cvs_branch = devel
variants:
- install:
type = kvm_install
mode = custom
install_command = "install_from_fedora_cvs.sh"
- ...
I used Fedora CVS as an example, but the user may use anything we can
imagine, to store KVM code (or pointer to its), possibly having
different branches.
--
Eduardo
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
