Re: Call for testers! (and updated pre-release tarball)

["David A. Desrosiers" <hacker gnu-designs com>]

> >         I had to grab autogen.sh from the gnome-pilot CVS tree to 
> > re-gen configure and aclocal to point to the right paths. It was 
> > missing the GNOME_COMPILE_WARNINGS macro definition

> I've added macros/compiler-flags.m4 to the gnome-pilot tarball.

	Same problem in -pre2 here. I'll see if I can narrow down why 
this is happening, but when I gen new config files and aclocal macros, 
I don't get the "undefined reference to `pi_buffer_new'" errors. 
pi-buffer.h is included (as is pi-source) in the right places, but its 
not catching the right paths, because aclocal insists on pointing it 
to the 0.11.8 version, even when I specifically point to the one in 
/usr/local.

	Generating the files anew with the cvs version of autogen.sh 
fixes the problem, configure works the same, and the errors with 
finding pi_buffer_free go away.

> > and there were some weird buglets with the rpath in the .la files 
> > in /usr/lib/gnome-pilot/conduits, which point to bogus, 
> > non-existant paths.

> ?? I'm not sure I follow, but do you mean that there were some 
> issues with overriding paths with ./configure switches?

	rpath gets clobbered, maybe due to the aclocal/m4 issues 
above, not sure as yet. If you grep the .la files in 
/usr/lib/gnome-pilot/conduits on an Ubuntu system (so far that I've 
tested), they contain bogus paths to pilot-link 0.11.8 which do not 
exist on the live system, and these paths come before the pisock link 
point in the final link, causing wonderful problems. It also overrides 
(excludes) any 0.12.0 libpisock lines and prefixes passed to the 
build.

	I had some issues last night with g-p just crashing wildly 
when I tried to connect over the network. I'm looking into it today, 
because I need to sync my calendar with my DAV server.


David A. Desrosiers
desrod gnu-designs com
http://gnu-designs.com