[oe] The truth about OE team being afraid of 'make' (was: Getting Started -Makefile)

[Chris Larson]

On Nov 21, 2007 6:28 PM, Michael 'Mickey' Lauer <mickey at vanille-media.de> wrote:
> Mike (mwester) wrote:
> > I ask the OE team why are they are philosophically opposed to driving
> > bitbake with make?
>
> Ok, seems now's a good time to tell you. Because that's where we came
> from back in 2002 -- from a Makefile-centric buildroot called
> OpenZaurus that didn't scale at all (which is inherent to the
> buildroot approach).
>
> We are so glad that by inventing OpenEmbedded and BitBake we have left
> this stoneage of building distributions behind.
>
> Getting back to the 'make' command reminds us of the painful past before we had
> OE.
>
> Then again, it could be just a paranoia of me, Chris Larson, and
> Holger Schurig ;)

GNU Make has some features now which it did not have then, that could
have made it slightly less painful to use for our work, but naturally
it wouldn't be worth going back to a more limited base now.  Back
then, it didn't support things like dependency on a file's existance
only, avoiding rebuilds when timestamps change, and there was no
'eval' to facilitate on the fly rule generation.  To summarize, make
just wasn't suitable back then, and it would be a great deal of work
to go back to it now, and it'd still be less flexible than what we
have today, and would suffer from portability headaches with make
versions.
-- 
Chris Larson - clarson at kergoth dot com
Dedicated Engineer - MontaVista - clarson at mvista dot com
Core Developer/Architect - TSLib, BitBake, OpenEmbedded, OpenZaurus