Make replacement for Erlang - suggestions?
Marc Ernst Eddy van Woerkom
Mon Sep 23 17:31:21 CEST 2002
> fundamental improvements. I will agree with you, the
> functionality of make is pretty much all there. I am
> sure there are not to many features that it is lacking
> at this time. I have two big problems with make
It has dependencies, lots of predefined targets (useful for
Java work) and the ability to do work via a call to the
> 1) The fact that you can not really write
> cross-platform make files.
I would argue that gnumake is available on virtually
every platform and that is not too hard to make stuff
work for many platforms under it.
Note that Ant has the same portability problem, once
you need to do work that is not in the predefined bag
> 2) The ugly non-intuitive syntax, and distribution of
> make files. (* This is of course my opinion, and I
> realize that not everyone shares this opinion with me.
I only hate the mix of different syntaxes between make
and usually bourne shell syntax.
> As for the first problem, for languages like C/C++ etc
> I don't really know how you could create a truly
> cross-platform makefile or at least a makefile without
> a lot of branches and tweeks for each platform.
The autotools (automake/autoconf/libtool) are not bad
either, they work one level higher, but have their
My ex-colleagues hated it because they were too lazy learn
the extra stuff and kept to their make/sh stuff despite
my autotools system being much powerful.
The manager wanted more than one person to be able to
support the build system so we had to agree on the lower
common denominator make/sh. :(
> I think make is so ubiquitous becuase its is available
> on allot of platforms, it's free, and replacing all
> it's functionality would take a significant amount of
> time. At this point it has a significant amount of
> market share as well.
What ever happened to the code sourcery attempt to replace
some of the build/configuration infrastructure?
They ran a competition which produced some interesting drafts
on what to create but which got never implemented, as far
as I know.
More information about the erlang-questions