Re: CHECK_STRUCTS/dmpstruct.h mechanism is broken.

[Achim Gratz]

Paul Eggert writes:
> On 4/11/19 12:55 PM, Achim Gratz wrote:
>> Yes, it has no autoconf, but
>> that's been replaced with the configure step from waf, which has exactly
>> one person you can ask if something doesn't work the way you think it
>> should.
>
> Yes, I considered waf too (coincidentally I was just talking to a
> student today about waf for one of his projects!). However, I don't
> think waf would be as good a match for Emacs, as it brings in more
> dependencies (Python, bzip2) and it's not yet clear that waf will be a
> stable-enough build tool for Emacs in the long run.

As a build tool it's OK I think, but the configury for Emacs would be
slow in waf, too, if it was implemented the same way as now.  That is,
unless you had a real expert for coding it up for waf in Python, which
would mean nobody else could understand what it does.

Meson would have the same problem (unless you sidestep and use autoconf
again) from what I understand of it.

> In contrast, if we simply use GNU Make we are reducing dependencies (by
> removing dependencies on autoconf/automake/m4) and we know GNU Make will
> be there in the long haul.

Again, I'd be more interested in autoconf picking up the obvious
opportunities for parallelization that are there for a start.  Then
Emacs would have to look at how the configuration process is organized
so it can be split into parts that are invariant and can be cached and
other parts that have to be re-run with option changes or always (for
whatever reason).  I'm pretty sure you'd be able to drop initial
configure time by a good factor on an SMP machine that way and
subsequent configure time to 10% of what it is now and that's be the end
of this discussion since any more speedup would not be noticeable
anymore.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Factory and User Sound Singles for Waldorf Blofeld:
http://Synth.Stromeko.net/Downloads.html#WaldorfSounds