[PATCHv2] build: allow some python variable overrides

[Jeremy Allison]

On Tue, Feb 11, 2014 at 08:15:20PM -0300, Gustavo Zacarias wrote:
> On 02/11/2014 03:09 PM, Andrew Bartlett wrote:
> > All I'm suggesting is that we should make it possible to discover and be
> > consistent? Can't this somehow be done via cross-answers, or a -- option
> > to configure?  I just really dislike magic environment variables.
> > 
> > Jeremy,
> > 
> > Can you give some guidance on what level of 'magic' should be acceptable
> > to impose on those cross-compiling Samba?
> 
> Hi Andrew.
> Doing it via cross-answers is possible, however normally the
> cross-answers would be basically static according to architecture/libc
> combination so modifying it on the fly wouldn't be the preferred method.
> A -- option would be great, environment variables are good as well.
> The acceptable 'magic' for buildroot is to try to avoid
> patching/diverging too much from upstream since it makes
> versions/security bumps possibly very complicated.
> Take a look at: http://patchwork.ozlabs.org/patch/319241/ explicitly the
> samba4.mk file to see how i'm building it at the moment.
> It still needs enhancements like possibly ads support.
> I can't speak for other build systems, others find patching like hell
> acceptable, in buildroot we aren't that many developers and it's all
> volunteer effort, hence the general attitude to avoid complicating
> things to much (if it can be described in env variables, -- options or
> similar is all good, that makes life easy).
> Eventually, i think, the idea would be to try reduce the cross answers
> file as much as possible (or even remove the need for it if that's
> possible), after all samba3 did build without it with just a couple of
> samba_cv_* defines that are pretty static for linux.
> I'll post more patches to try to simplify the cache soon.

Thanks ! In the meantime I've pushed your fix with
David and my 'Reviewed-by'. Thanks a *LOT* for your
help on the cross compilation problem.

Jeremy.