This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: configure: error: invalid package name: extra-includes


Gerrit wrote:

> Peter wrote:
> 
> >> The last time I tried to build ggi-project failed with compilation
> >> errors.  That probably means that I was able to configure it 
> >> or parts of
> >> it.
> 
> > I have spent some time on getting libggi working. I believe 
> it does work
> > now, you have to use the cvs version though as the released 
> versions do
> > not have all the patches that has been commited lately.
> 
> > The problem is not that I can't get it to work. It works 
> *most* of the
> > time. *Sometimes* there's this mysterious failure though. 
> I'm tired of
> > adding spaces to the configure command and I would also like
> > "./config.status --recheck" to work reliably.
> 
> > What does an extra space between configure arguments really 
> change? This
> > is friggin' strange.
> 
> >> I always update the autogenerated part when a package is going to
> >> build shared libraries, try: autoreconf --install --force 
> --verbose in
> >> the source directory.  Well, you do this anyway when building 
> >> from CVS?
> 
> > No, there is an autogen.sh script included though. It 
> basically boils
> > down to this:
> > aclocal
> > autoheader
> > automake --add-missing
> > autoconf
> 
> > Anyway, I tried:
> > WANT_LIBTOOL_VER=1.5.6 WANT_AUTOMAKE_VER=1.8.5 
> WANT_AUTOCONF_VER=2.59
> > autoreconf --install --force --verbose
> > and I can still reproduce the bug. I can configure with the 
> "correct"
> > number of extra spaces though, but it later bugs out during 
> make so that
> > is a definite step backwards.
> 
> >> Which version of configure, automake, libtool are you using?
> 
> > autoconf 2.59 and automake 1.8.5 and an old libtool 
> (libtool is included
> > in libgii, I've had to throw quite a few patches at it to 
> make it work
> > with cygwin/mingw. I reported a dlltool/dllwrap problem to this list
> > earlier related to this and got the advise to use "gcc -shared").
> 
> autoreconf -force -install also  installs the new libtool, very good
> idea to do it, since patching the included libtool is error prone.

It might be a good idea in general, but not in this case IMO. I'm no
libtool guru though, so the build problem autoreconf causes might be
easy to fix.

The reason I have patched libtool is that the ggi maintainers are
waiting until after their next release to bring in the 1.5 series of
libtool. I decided not to wait for that.

> I'll try to build the latest release, where can I find a summary about
> all the recent patches / fixes?

I don't think there's a summary on the level you'd be interested in,
other than the cvs log of these files:
http://cvs.sf.net/viewcvs.py/ggi/ggi-core/libgii/m4/libtool.m4
http://cvs.sf.net/viewcvs.py/ggi/ggi-core/libgii/ltmain.sh
Look for commits by me, or commits by cegger referencing me as the
originator. Hmm, when scanning through that list, it looks as if most of
my changes are mingw related. The bottom line is that I have not kept
track of the changes.

It is entirely possible that the changes that has been made to libtool
are bogus. It is, as you say, error prone to fiddle with libtool.
Especially when you, as is the case for me, don't really know what
you're doing and are just making educated guesses. I did make it work
though.

There has also been changes made to the configure.in file
http://cvs.sf.net/viewcvs.py/ggi/ggi-core/libgii/configure.in

Most changes have been made to other parts of the tree but as I see it
those can't be related to this script argument parsing bug.

Otherwise, these pages list high level changes:
http://www.ggi-project.org/packages/libgii.html
http://www.ggi-project.org/packages/libggi.html

Here are some build instruction if you are going to go for the build or
gii/ggi. You will not need the DirectX SDK to reproduce the bug, do
however include the --with-extra-includes option, it is a good way to
trigger the bug. The bug can be triggered with other options as well
though.
http://cvs.sf.net/viewcvs.py/ggi/ggi-core/libgii/doc/README.directx?view
=markup
http://cvs.sf.net/viewcvs.py/ggi/ggi-core/libggi/doc/README.directx?view
=markup

All of the above is perhaps too much info to sift through and if you ask
me, blaming autotools is a red herring. The generated configure script
looks ok to me, and to further support that claim I have tested my
previosly provided steps to reproduce on Solaris, AIX, Linux and mingw
and have not been able to trigger the bug. That is, I have moved the
configure script that has been generated on and fails on cygwin to those
platforms and there is no problem. On the other hand, I can easily
reproduce on cygwin; that usually happens on the 1st, 2nd or sometimes
3rd attempt. Therefore, if you ask me, the problem is in script
execution/invokation, not in script generation.

Cheers,
Peter


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]