Re: gnustep-make experiment

[Nicola Pero]

Thanks for your suggestions

You seem to think that the current solution lacks:

 1. a way to compile without setting GNUSTEP_MAKEFILES

 2. a way to get the GNUstep bundle/libs/tools search paths for use in 
configure scripts etc.

I agree we can do better to address those.  Solutions that came to mind as 
natural ones would be to --

 1. have a (trivial) makefile fragment that locates gnustep-make and you can 
include in your project to make it bootstrap without having to rely on 
GNUSTEP_MAKEFILES being set (I suppose we should also
consider just putting it into your make include path).

 2. have a (trivial) shell script that can output the list of GNUstep 
bundle/libs/tools search paths.  We install this as a system tool, so under 
Linux FHS it will automatically be in your PATH and you can just invoke it 
whenever you need (similar to the all the config files of other software that
you mention).

Maybe for 2. we could take the standard OpenStep search methods (for bundles 
etc.) and have shell script equivalents for them.  That would seem like a 
reasonable way of designing our shell 'API' to get the filesystem locations, in 
a way that would work with Linux FHS as well.

I was now feeling like jumping in and implementing all that, but it will likely 
be a few
iterations and discussions before we get something good and stable, and if 
we're trying 
to make a release in the next few weeks, it's not the right time to add new 
stuff.

I'd rather spend some time documenting what we already have before we try 
working on the next 
steps (nobody seems to have a clue about all the new stuff in gnustep-make).  
So can we come 
back to this in a few weeks ? ;-)

Thanks