From kde-core-devel  Tue Jun 14 12:03:49 2005
From: Stefan Teleman <steleman () nyc ! rr ! com>
Date: Tue, 14 Jun 2005 12:03:49 +0000
To: kde-core-devel
Subject: Re: Build system for KDE4
Message-Id: <200506140803.49870.steleman () nyc ! rr ! com>
X-MARC-Message: https://marc.info/?l=kde-core-devel&m=111875121732189

On Monday 13 June 2005 20:37, Michael Pyne wrote:
> I once used a plain 'ol Makefile with a Perl script to install
> icons for my program abakus.  It sucked less. :P

> I'm sick and tired of having to suffer through all the bloated
> layers of autoconf, automake and libtool just so that someone can
> fail to build KDE on UnixWare somewhere.

Well then it looks like this is a really well thought out plan.

first, we have a generically vague statement "automake sucks". i have 
several of these too: Python sucks, C++ template function pointer 
syntax sucks. are they useful ?

then, we have several anecdotal reports of individuals having built 
one part or another of KDE on their machine (sometimes just one 
program) "and it worked really well". uhhm, yeah. i once caught a 
fish *This Big*.

sure, let's make the KDE build process a separate large scale software 
development project. "It's not only a desktop, it's a build system 
too!" And while we're at it, let's rewrite GNU make from scratch, in 
Perl and Python. because the current GNU make "sucks". that's what 
users really want in KDE. a Build Challenge.

see, in the world i come from, (commercial software), you can't do 
this (well you can, then you get fired). if you really want to change 
the build system for a 5+ million LOC project, first you make a case 
for it by explaining the benefits of the *new* build system, and not 
by repeating "the old system sucks" to everyone else involved, 
because that is not news.  then, you prove to everyone else that it 
can be done, and it's worth the effort, by building KDE with SCons, 
and making your SCons build system availale to everyone, to test, and 
duplicate your spectacular results. *then* you compare the two build 
systems. *then* you decide if it's really worth the hassle and the 
risk. which at this point i can already say it isn't, because the 
only currently documented advantage of SCons over auto* is that its' 
"less mature and it needs a lot of work".

on the other hand, commercial software is NoTkEwL. it's based on this 
old fashioned idea of "Implementation Trumps Theory". 

just my 0.02.

--Stefan

-- 
Stefan Teleman          'Nobody Expects the Spanish Inquisition'
steleman@nyc.rr.com                          -Monty Python