Subject: Re: Build system
From: Eric W. Sink (eric@sourcegear.com)
Date: Sat Jun 17 2000 - 16:44:54 CDT
>Building makefiles for you IS one of the things that autoconf does.
#begin turbo-Luddite-mode
Yes, yes, I know... But the makefiles it generates will *not* solve
all of the problems which have been carefully addressed with the
current system. In order to get a 100% replacement for the current
build system, we'll have to re-do the work on the autoconf-generated
setup.
In the general, the pattern looks like this:
1. Programmer gets bored or frustrated with the effort of
maintaining some piece of mature code.
2. Programmer finds a solution which promises to solve the
whole problem automatically and effortlessly.
3. The new solution fails to deliver all of the functionality,
features or stability of the previous solution.
4. Programmer begins fixing all of the minor problems. The
process takes longer than expected.
5. Go back to step one.
Note that nothing in the process above has produced any benefit at
all which is visible to *users*.
I call this process "Developer Churn". :-)
BTW, the loop is usually not infinite. After enough iterations,
it spontaneously exits down to step #6:
6. Programmer realizes that all of the following are
more-or-less synonymns:
mature code
old code
working code
ugly code
code which is difficult to maintain
stable code
good code
Forgive my crabbiness, but when I founded this project, there were
several very wise people who insisted that a community-developed
desktop GUI app could *never* succeed. They guaranteed me that
it could not work, because 'the project will always attract
people who are driven by the enjoyment of the coding and development
process, not people who are driven by the enjoyment of building
things which meet the real needs of actual users'.
(There is nothing wrong with "people who are driven by the
enjoyment of the coding and development process", BTW. The
assertion that was being made was simply that most of the
developers on community projects are usually coding to
"scratch an itch", and that is the prevailing explanation
for why most of the successful open source projects do NOT
have a consumer grade user interface.)
Anyway, I'm still carrying a torch for AbiWord as the project
which will prove them wrong.
#end turbo-Luddite-mode
If my advocacy on behalf of users ever causes offense, please
forgive.
-- Eric motto -- Users rule and nerds drool
This archive was generated by hypermail 2b25 : Sat Jun 17 2000 - 16:40:25 CDT