apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Thom May <t...@planetarytramp.net>
Subject Re: cvs commit: apr/test Makefile.in
Date Wed, 14 Aug 2002 20:27:11 GMT
* Greg Stein (gstein@lyra.org) wrote :
> On Wed, Aug 14, 2002 at 12:11:05PM -0700, Justin Erenkrantz wrote:
> >...
> > My concern is that this is attempting to enforce a versioning
> > scheme that isn't what everyone uses.  It's what you would use
> > if you use /usr and want to support multiple versions installed at
> > the same time.  Greg's commit locks everyone into this model, and the
> > more I think about it, I think that is harmful.
> It is an improvement on our existing situation. In that sense, it is not
> harmful. The previous situation was hella more harmful.
Definitely agreed.
> >...
> > If APR supported config.layout (easy enough to add), a layout
> > could have:
> I would totally support adding such a notion, along with the
> ${major_version} example that you provided.

This is really easy to do - the debian packages were doing exactly that for
a long time. I can resurrect that patch and apply it this evening if people
think it's a positive step?

> Note that I won't hold 0.9.0 for that, though. :-)  IMO, if the more
> flexible installation appears in 0.9.1 ... we're still doing fine by our
> users.
> Do you suppose that the layout processing could be app-independent? For
> example, APR provides the functionality, and APR, httpd, and SVN could all
> use it?
> [ I'm not trying to add work; if you do an APR-only solution, I'm still fine
>   with that... just another add'l feature ]
For sure. the layout stuff is defined as an autoconf macro, so svn would
just need to sinclude apr_layout.m4 or whatever, and then call the macro.
We would, I guess, update httpd so we just had one canonical
APR_LAYOUT implementation.

Thom May -> thom@planetarytramp.net

* moshez tries to understand why in god's name /usr is symlinked to / on the hurd
<vorlon-work> moshez: the same reason that /body/head is symlinked to /body/ass in some
Hurd developers?
<vorlon-work> No, that's unkind.  /usr -> / makes more sense than the anatomy of
said developers.

View raw message