My newbie detailed analysis of the profile

Hendrik Volker Brunn hvb at gmx.net
Thu Sep 7 12:56:20 PDT 2000


On Thu, Sep 07, 2000 at 03:28:55PM -0400, Gerard Beekmans wrote:
> 
> Am I correct is I say that you are suggestion we create a better make program 
> to parse Makefile files? I think I'm completely missing the point now, so 
> please elaborate if you don't mind.

He probably means we're just building a top-level-makefile. And he has
some point there. The backend we're designing is to be implemented in
whatever language we dream of, but - and that's an big point - it will
always come down to 
./configure
../make
../make install
aand these will be called by the backend. So there is little point
in abstracting our profile. We break it apart in little pieces and all the
backends will put it together to the exactly same commandline that is run
from inside perl, python, a compiled c-program. I can see no reasons to
take shell commands and rip them apart in their smallest pieces, let them
being parsed by an backend that itself will issue the shell commands we
had before.

So why should one code a binary that will run shell commands? We will rely
on configure, make, gcc ... anyway. So I believe we should save these
efforts - you won't get me with "because we can", I'm living this way
since I know Linux - and take the nearest possible approach - a Makefile.


Just some thoughts.

And now: throw your stones!

hvb

And remember Gerard said we shouldn't kill each other while he's away.

-- 
Linux, of coures, cannot use such excuses because under Linux - 
"if something is possible in principle, then it is already
implemented or somebody is working on it".





More information about the alfs-discuss mailing list