Componentizing configure.in?

Bill Gribble grib@linuxdevel.com
03 Dec 2001 09:40:18 -0600


On Mon, 2001-12-03 at 09:04, Derek Atkins wrote:
> Is it really that bit a deal?  Most normal users should be using the
> "stable" distribution which has already been autogen'ed.  Is there any
> particular reason you can't autogen the code on one system and then
> copy the sources to another?  Yes, I realize that that, too, is a
> PitA.

I'm aware this isn't an issue for normal users (who don't compile at
all, BTW :). It is a real problem for me, and it's potentially an issue
for anyone who wants to use components of the gnucash source (like the
gnc-module system or the engine) without having to install all the GNOME
dependencies and other things that make gnucash so famously a pain to
build. 

I am trying to compile on a machine with a completely different
architecture and filesystem layout, and different versions of many tools
and libraries, so I'd prefer to not try to just autogen on one machine
and move the tree.  It would probably work but it would not be the right
way to solve the problem, especially considering that I have a CVS tree
on the target machine and want to be able to update it and rebuild
without having to involve another machine. 
 
> If you go the modularization route, my only request is that
>    ./autogen.sh
> defaults to pulling in everything.

That was what I suggested. 

b.g.