Re: trying to track down memo-file conduit bug

[Matt Davey <matt corvil com>]

> I've attached a patch for the gnome-pilot-conduits file
> memo_file/memo_file_conduit.c. I basically changed the load_records
> function (and function call) back to how it was before 2.0.10. JP
> changed load_records to prevent the memo_file conduit from crashing
> since he didn't have time to fix it. I left the rest of the file alone,
> and it works now. :)

That's great!

> I had to make sure that gpmarshal.h was included in the
> gpilotd/gnome-pilot-conduit-standard-abs.h file.
> 
> I also had to make sure that gpmarshal.h was copied over to the
> /usr/include/gpilotd directory so I could rebuild the
> gnome-pilot-conduits package (this meant modifying the
> gpilotd/Makefile.am and gpilotd/Makefile.in files - I've included those
> as attachments too, in case someone cares).

This is my mistake.  Since my changes, gpmarshal.h is required by
gnome-pilot-conduit-standard-abs.c
I added the #include to gnome-pilot-conduit-standard-abs.h, not realising
that this file is a public header file.  The fix is simply to do the
#include from inside gnome-pilot-conduit-standard-abs.c (as is currently
done by gnome-pilot-conduit-sync-abs.c).  Then we won't have to
worry about making gpmarshal.c externally visible.


Matt