Re: Squid-2.5

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Wed, 03 Oct 2001 17:43:18 +0200

Robert Collins wrote:

> Granted... but we already filter the autotools created files. IMO this
> setting is another user convenience. One of the notes in
> the --enable-maintainer-mode is that a downside is the developers no
> longer experience what the user does.

Good point.

> Sure. We manage to muck up a release and haven't bootstrapped it because
> this switch has disabled the built-in protection against out-of-date
> files and we are not yet generating the releases via 'make dist'. Every
> squid user then writes us to complain that option xyz never configures -
> because autoconf.h.in was out of date.

Which is best fixed at the distribution point I think. We will get even
more complaints if the "automake" rules triggers on each attempted
build..

> bootstrap.sh should not be in the distributed tarballs. It's a developer
> only tool - like --enable-maintainer-mode - that is _only_ required when
> downloading from a CVS without the autotool generated *.in and configure
> files.

There is also other developers not using CVS, and having the
bootstrap.sh script makes life considerably easier for them.

> Ah. Did you use the missing script from automake, or a custom brewed
> recipe? This _can_ be tricky.

A custom brewed thing. But still, I see it as likely that the automake
scripts will fail. What happens for example if the users clock happens
to be running a couple of months/years late?

> I understand - I'm happy from my user/developer perch right here - just
> raising the issues as I see em. I'd hate for us to make things harder
> than they have to be.

Good.

--
Henrik
Received on Wed Oct 03 2001 - 09:42:29 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:24 MST