gpsd-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: would it be time to replace scons with meson build system?


From: Eric S. Raymond
Subject: Re: would it be time to replace scons with meson build system?
Date: Wed, 5 Feb 2020 00:18:37 -0500
User-agent: Mutt/1.10.1 (2018-07-13)

Hal Murray <address@hidden>:
> 
> address@hidden said:
> > No.  I went scons to get *away* from autoconf.  I hate it with a fiery,
> > burning passion. 
> 
> Would you please say a bit more.  What's wrong with it?  If you were starting 
> over, what would you use and why?

It's hideously complex, with lots of moving parts.

It is absurdly difficult to troubleshoot recipe problems because of
the two-phase architecture - the errors are thrown from generated
Makefiles at points where the relationship to the sourvcec recipe is
not obvious.

The documentation is a frustrating combination of exhaustively detailed
with unhelpful. Every twig on every tree is explained, but an overview
conveying generaive knowledge is difficult to put together.

If you ever try to gain insight into autconf's behavior by studying
its code, you'll find it is hacks and kludges all the way down,
fractally ugly and overgrown.

It spends a lot of its complexity solving yesterday's problms. The
success of C and Unix API standardization means that most of its
configuration apparatus has become useless.

Starting over...

Definitely a single-phase system like scons.  But, today, not
scons - I think that was a good choice when we made it, but
scons development has stalled and stagnated.

waf has served NTPsec well, but its documentation is awful.
I'd be willing to look at alternatives like meson, cmake,
bitbake, buildout, or gradle.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>





reply via email to

[Prev in Thread] Current Thread [Next in Thread]