At Thursday 22/06/2006 23:49 +0200, Claus-Justus Heine wrote:
Please use the stuff in CVS, not that "alpha" snapshot.
The reason why we are using that alpha snapshot (and people are
building rpms from it...) is because you have to run ./reconf on
geomview CVS to generate a release by going through all the
autoconf/automake stuff. This presumes you have a lot of extra stuff
installed. This presumes that you know to run reconf - it's not documented.
I could not find a system on which I can create a snapshot from cvs
that a mere mortal like myself can make and install -- and I don't
trust my cygwin setup enough to generate a good configure file.
Although the reconf process completes these days, I still get way too
many warnings from reconf -- below at end. (As it stands, I can't
compile any geomview on cygwin at the moment anyway -- the cygwin
guys broke d_ino support, which is allegedly optional under POSIX,
but relied on by src/bin/animate/glob.c.)
I would be more than happy to replace that alpha snapshot with a
later snapshot -- if someone will provide me a snapshot they believe
can be configured and compiled, already reconf'd with Makefiles on a
system where autoconf is a known and trusted quantity.
Until then, that alpha snapshot is available as the last time I was
given something that was trusted to be built.
Actually, we really should make a
"release", just to stop people using that "alpha" snapshot beast.
Yes. No argument here. I've been saying that for several years.
I'm using (occasionally) geomview on X86 and X86_64, and once upon a
time I was using it
on an Alpha Workstation (R.I.P.). I'm pretty sure that most of the
stuff should work. If I
find the time (unlikely) I can try to run SAVI on my X86_64 next weekend.
SaVi should be 64-bit clean, thanks to your previous patches, which
I've incorporated. But this is about Geomview, not SaVi -- SaVi is
trivial, and its problems are trivial, by comparison.
The crashes might be caused by insane usage of variable argument
functions inside Geomview
(i.e. that fct(foo, ....) stuff). The original version of Geomview
liked to do some fancy
things with the va_args type which is -- so to say -- a super-opaque
data type;
it's how you pass a variable number of args to printf etc., so
everyone uses it, probably unknowingly.
this
should be fixed in the CVS archives (but _NOT_ in that unfortunate
"alpha" snapshot, BTW,
ever gave a thought to the meaning of "alpha" in relation to
software development? An
"alpha" snapshot is no better (i.e. no more stable) than the HEAD
revision of any CVS
archive.
Which is why I clearly labelled it as such. No quotes needed. It's alpha.
Please see above for reasons for distributing the Geomview alpha
snapshot. At least I titled it with fair warning. Given the problems
Geomview 1.8.1 has with recent compilers and its stream buffer
problem, the alpha snapshot is still imopreferable to 1.8.1.
At least the people preparing "distributions" should be aware of that fact.
If they can't use 1.8.1, what other choice do they have? They're not
going to pull from CVS and put that in a distro.
Just
to prevent flame-wars: that va_list stuff is really somewhat
difficult, the above meant no
offend to the Geomview developers; at least I needed some time to
figure out how to code
the stuff intended by the Geomview developers correctly. RTF glibc manual).
JFYI: I'm using a quite recent Gentoo distro, and gcc-4.1.x, but
Geomview was working (for
me) also with gcc-3.x, that should not matter. There are problems
with either openmotif or
lesstif
exactly. the debian bug reports lesstif. Other users report
openmotif. I willing to even believe that they've got common code and
similar 64-bit problems, and that the lesstif guys are addressing
their 64-bit problems.
; at least my xemacs does not work stable with any Motif clone
(either on X86 or
X86_64). Could also be an xemacs bug, of course.
Give the CVS stuff a try! I'm pretty sure it will work. The
sourceforge pages have
detailed information about how to download CVS archives.
I just did that.
$ ./configure -with-motif=/usr/X11R6/ -with-opengl=/usr/X11R6
bash: ./configure: No such file or directory
At that point. most people will give up - running reconf isn't even
documented in a README in CVS! It's black magic to stop people from
building Geomview!
See above for why CVS is not useful to normal human beings. I use CVS
all the time with SaVi -- unlike Geomview, that configures and builds
cleanly from CVS head, but then it doesn't use autoconf, and the
final Makefiles are actually in CVS.
CJ, do you want to provide a new reconf'd alpha snapshot?
thanks,
L.
$ ./reconf
+ test '' = -n
+ rm -f config.cache
+ aclocal -I m4
m4/lf_local.m4:33: warning: underquoted definition of LF_LINK_HEADERS
run info '(automake)Extending aclocal'
or see http://sources.redhat.com/automake/automake.html#Extending-aclocal
m4/lf_local.m4:85: warning: underquoted definition of LF_SET_INCLUDES
m4/geom_local.m4:23: warning: underquoted definition of GEOM_FIND_L_OPTION
m4/geom_local.m4:66: warning: underquoted definition of GEOM_FIND_I_OPTION
m4/geom_local.m4:110: warning: underquoted definition of GEOM_FIND_LIBC_VERSION
m4/geom_local.m4:130: warning: underquoted definition of GEOM_AC_PROG_CXX_WORKS
m4/geom_local.m4:142: warning: underquoted definition of GEOM_AC_PROG_CXX
m4/geom_local.m4:187: warning: underquoted definition of GEOM_GET_CACHED_VALUE
m4/geom_local.m4:203: warning: underquoted definition of GEOM_REQUIRE_XFORMS
m4/geom_local.m4:231: warning: underquoted definition of GEOM_CHECK_XFORMS
configure.in:679: warning: AC_CONFIG_SUBDIRS: you should use literals
autoconf/status.m4:1077: AC_CONFIG_SUBDIRS is expanded from...
configure.in:679: the top level
+ autoconf
configure.in:679: warning: AC_CONFIG_SUBDIRS: you should use literals
autoconf/status.m4:1077: AC_CONFIG_SUBDIRS is expanded from...
configure.in:679: the top level
+ autoheader
autoheader-2.5x: WARNING: Using auxiliary files such as `acconfig.h',
`config.h.bot'
autoheader-2.5x: WARNING: and `config.h.top', to define templates for
`config.h.in'
autoheader-2.5x: WARNING: is deprecated and discouraged.
autoheader-2.5x:
autoheader-2.5x: WARNING: Using the third argument of `AC_DEFINE' and
autoheader-2.5x: WARNING: `AC_DEFINE_UNQUOTED' allows to define a
template without
autoheader-2.5x: WARNING: `acconfig.h':
autoheader-2.5x:
autoheader-2.5x: WARNING: AC_DEFINE([NEED_FUNC_MAIN], 1,
autoheader-2.5x: [Define if a function `main' is needed.])
autoheader-2.5x:
autoheader-2.5x: WARNING: More sophisticated templates can also be
produced, see the
autoheader-2.5x: WARNING: documentation.
configure.in:679: warning: AC_CONFIG_SUBDIRS: you should use literals
autoconf/status.m4:1077: AC_CONFIG_SUBDIRS is expanded from...
configure.in:679: the top level
+ automake -a
configure.in:679: warning: AC_CONFIG_SUBDIRS: you should use literals
autoconf/status.m4:1077: AC_CONFIG_SUBDIRS is expanded from...
configure.in:679: the top level
configure.in: installing `./install-sh'
configure.in: installing `./missing'
doc/Makefile.am:11: installing `doc/texinfo.tex'
src/bin/animate/Makefile.am: installing `./depcomp'
configure.in:3: installing `./config.guess'
configure.in:3: installing `./config.sub'
+ test '' '!=' no
+ ./configure
Claus-Justus Heine <[EMAIL PROTECTED]>
Ftape - the Linux Floppy Tape Project http://ftape.dot-heine.de/
NFS Swapping for
Linux http://nfs-swap.dot/heine.de/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
