fate of Makefile.am

[David Faure]

On Wednesday 12 October 2005 19:42, Leo Savernik wrote:
> Hence all the Makefile.ams would stay as is 
> (minus Makefile-specific stuff), and the generated SConscripts become the 
> Makefile.in/Makefile equivalent.

I disagree with this idea, even though I initially was thinking about that too.

We would lose most benefits of using scons. For instance if you create 50
test programs in a directory (like kdelibs/kdecore/tests), with scons you
can write a loop, and be done in 6 lines. With Makefile.am you needed 100 lines.
So I don't really want to keep the Makefile.am syntax as an input to scons,
it's much less powerful.

It also reminds me of when I had to use a "pascal to C" compiler to compile
pascal code. Every error message would be in the generated C code, and
was completely impossible to understand since I didn't write that code and
it didn't look at all like the initial pascal code.
It's the same here; with automake it was hard enough to debug problems,
but if now Makefile.am is converted to python and you get incomprehensible
python error or behavior, it will be really a mess.

-- 
David Faure -- faure at kde.org, dfaure at klaralvdalens-datakonsult.se
KDE/KOffice developer, Qt consultancy projects
Klarälvdalens Datakonsult AB, Platform-independent software solutions