Subject: Re: gdb
From: kragen@pobox.com (Kragen)
Date: Mon, 15 Dec 1997 13:39:12 -0500 (EST)

On Sun, 14 Dec 1997, Craig Burley wrote:
> I agree.  However, I believe this is a temporary phenomenon.  Proprietary
> software was once abysmal, even worse than today's freed software,
> in terms of packaging issues such as those discussed here. 

And, as someone pointed out regarding Oracle, still is in some cases.

> E.g. clearly the organized help shepherding successful
> products like gcc and Linux (rms and Torvalds) know better
> than to let "just any old code" in that might break the
> products' functioning.
> 
> I believe there isn't any serious difference between that and
> people in the same position not letting in "just any old
> feature/documentation/packaging" that might break the products'
> proper overall packaging and ease of use.

RedHat, which is selling to the PC user market, has put a high priority on
this with its RPM.  A lot of other people are using RPM now, because they
recognize the value of good packaging.  (Information on the goals and
features of RPM can be found at and near
http://www.rpm.org/support/RPM-HOWTO-1.html.)

qmail is distributed under a licence whose sole purpose is to maintain
good packaging.  You're only allowed to distribute the original qmail
archive unless the author approves your distribution; his criteria for
doing so are explicitly concerned with packaging quality.  See
http://www.pobox.com/~qmail/dist.html for more details.

> much as we do those that help validate code.  I notice
> there's no equivalent to "gcc" (or any decent C compiler)
> for makefiles, shell scripts, and so on, *and* that lots
> of the hair in dealing with a highly complicated *program*
> like gcc is in these overlooked areas.  

You mean, a tool that pores over makefiles and shell scripts and tells you
why they're unportable?  I'm not clear on what you mean by "an equivalent
to any decent C compiler".

Kragen