Re: Building CVS


Subject: Re: Building CVS
From: Paul Rohr (paul@abisource.com)
Date: Fri Aug 10 2001 - 15:02:14 CDT


At 09:05 PM 8/10/01 +0200, Hubert Figuiere wrote:
>You did run the wring configure script. Dom talked about wv's, not abi's...
>
>In abi, do a make realclean. Then you'll recover the diving make that
works[*].
>In wv, make distclean.

Yep. In the old-style diving make environment, doing the following should
always work:

  cd work/abi (or wherever you keep our trees)
  make realclean
  make

That's because abi's diving make in turn triggers the appropriate level of
cleanliness for *all* its peer modules. In the case of wv, the rule gets
translated to a "make distclean" over there.

As Dom suggested, once you know how this works, you can get away with
cleaning fewer trees (roughly) as follows:

  cd work/wv
  make distclean
  cd ../abi
  make

>Sorry for the confusion.
>
>Hub
>
>[*] the autconf stuff works, but not the way you expect.

Ditto. Things were much less confusing when we only had one way to build
things.

Since there were no autoconf experts on the original team, we built
an XP diving make system instead. I'm slowly being convinced that an XP
autoconf-only build environment *might* be feasible, but the following
populations are still 100% disjoint AFAICT:

  - autoconf experts
  - people maintaining all the non-Unix builds

Until those two populations cross-pollinate enough, we'll have difficulty
bridging the gap.

Speaking only for myself, it still takes me *much* less time to update a few
simple diving makefiles than it does to learn enough autoconf stuff to jump
the gap. Thus my continual insistence that folks (like Sam) try to jump
that gap from the other side, instead. :-)

Paul
motto -- makefiles are direct & simple, autoconf is complex & powerful



This archive was generated by hypermail 2b25 : Fri Aug 10 2001 - 14:54:33 CDT