🚀 go-pugleaf

> In article <STEINARB.92Oct4172851@hagtorn.idt.unit.no>, steinarb@idt.unit.no (Steinar Bang) writes:

>> One of the files in the docs/notes directory of mach.cs.cmu.edu
>> mentions that the Mach people at CMU are working on moving the
>> Makefiles over to a format usable by a freely available Make (The OSF
>> ODE Make based on the BSD-Reno Make from the).
>>	...
>> Do the changes necessary to the CMU Makefiles to make them work
>>     with the ODE Make (or the GNU Make for that matter), and nicely
>>     ask CMU to start using these Makefiles in their distribution.

>> Suggestion 2 is the most desirable solution. It is however a biiig
>> job. It is not "fun" enough that anyone ever is likely to sit down and
>> do it for free, which means that some company or organization will
>> have to pay someone to do the work.

>> The question: is who is going to do it? CMU don't have any incentives
>> for the time being. OSF? FSF? I don't know.

>>>>> david@postman.gr.osf.org (David George) writes:

> OSF RI has already done this work for the Mach 3.0 (NORMA) microkernel
> and OSF/1 single server, we use ODE as our day to day build environment but
> generally to build for i386 AT architectures (AT, COMPAQ SYSPRO, COROLLARY).
> This work is available to OSF/1.0 licencees.

But it is not available to non-licensees, so it will not be any use to
FSF or people at non-BSD licensee organizations (or hobbyists wishing
to roll out eg. GNU Hurd on their home computer). I am in the work of
getting the BSD licensing stuff fixed, so hopefully this will not be a
problem for me personally much longer.

It will still be a problem for the FSF and experimenters at
organizations without a BSD license, though.

> I can appreciate the amount of work required to convert all the
> platforms CMU build to the ODE environment [Mary Thompson detailed
> some of the problems she'd encountered in her Email to you of the
> 10/01].

Yes. It is because of that email message I am bringing this issue up
in a gnu newsgroup.

> Extensions to ODE are needed for some platforms and cross
> compilation is not supported with the version we use.  I can't
> realistically see CMU wanting to support two build environments
> while they convert.

Neither can I, really. So maybe the best solutions for all parties
would be to make GNU Make support the CMU enhancements.

*Even* though this might mean cluttering GNU Make with facilities
needed to compile a single program (ie. the Mach 3.0 microkernel).

I would be curious to know the opinions of Roland McGrath (GNU Make
maintainer) and Michael I. Bushnell (GNU Hurd developer) on this
subject. 

- Steinar