From: Godmar Back (gback@marker.cs.utah.edu)
Date: Sun Jan 10 1999 - 15:57:15 EST
On the same linux box where it worked for unix-jthreads, it now fails for
oskit-pthreads. I do not see where the rule for libkthread.la is.
I do not understand why it worked before for unix-jthreads.
make[3]: *** No rule to make target `systems/oskit-pthreads/libkthread.la', needed by `libkaffevm.la'. Stop.
Alexandre, here's my suggestion:
Take until Wednesday or so to bring the kaffe cvs repository in a working
state. I would like to define `working' as:
On *all* systems where the cmd sequence
"cvs update; ./configure; make all install"
produced a running and installed kaffe executable *before* the
automake conversion, it will still do so *after* the conversion.
In my opinion, that's the least we should expect.
Note that my definition of `working' is independent of what version
of autoconf/automake/libtool/cvs whatever is installed on these systems.
I will be forthcoming in that I restrict the definition of `working' to
exclude all systems that do not have gmake installed, although that's
a step back too since I was able to use BSD's make before the conversion.
If you're not able to bring the cvs repository in a working state, I would
like you to unroll all automake related changes by Wednesday. I am opposed
to making Kaffe's build environment a playground for alpha-testing some
half-baked tools. Depending on future, as of yet unreleased versions of
software (and blaming existing and widely used software while at the same
time not providing work-arounds, which is how I interpreted the cvs update
time stamp problem) is quite unacceptable in my opinion.
I recognize the problems that libtool solves, namely the static/dynamic
lib issue, and I think that using it to solve these problems is a good
thing in principle.
What do you and others think?
- Godmar
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:34 EDT