[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[avr-libc-dev] [RFC] multilib
From: |
Theodore A. Roth |
Subject: |
[avr-libc-dev] [RFC] multilib |
Date: |
Mon, 15 Sep 2003 16:00:00 -0700 (PDT) |
I've been thinking about this for a while now and the recent problems
with people seeming to not get everything to install have accelerated
things...
The use of the multilib build system doesn't seem to be all that
useful to me. I don't see the point of building and installing
avr-libc for only one arch, e.g. avr3. Does anyone even do that?
Besides, I've never liked that fact that the build system always puts
the avr[345] dirs in the top src dir instead of into the build dir or
where ever configure was run from.
Also, the multilib support (via config-ml.in) seems to make things
more complicated than needed.
I've been thinking about ripping the multilib support out when I do
the convertion to the new autotools.
So, what I want to happen is the following:
- Remove doconf, domake, dodist, and rename reconf to
bootstrap. These don't fall in line with the common './bootstrap
&& ./configure && make && make install' idiom used by other tools.
I've never used the doconf, domake, dodist scripts since when I
see configure script, I assume that it is there for me to use to
configure the package (this seems to be what those who had the
recent problems thought too).
- 'mkdir obj && cd obj && ../configure && ...' should be the
preferred method (rename obj to anything you like) or building in
the top src dir, but building in top src dir should work.
- The avr[345] obj files should end up in the directory that
configure was run from under the avr[345] subdirs. Thus, the
avr[345] subdirs will always be under $(top_builddir), thus
localizing them to a specific run of configure.
- convert to autoconf-2.57 and automake-1.7.6
I'd like to start this next week since we are early in the 1.1 dev
cycle and it's bound to break something until it gets pounded on for a
while.
Ted Roth
- [avr-libc-dev] [RFC] multilib,
Theodore A. Roth <=