your mail

[tom mystery antar com (tom)]

   Toshio Kuratomi wrote:

 >> with: 
 >> ** ERROR **: BadMatch (invalid parameter attributes)
 >> IOT trap/Abort
 >> 
 TK> You need to get an updated version of imlib -- either from the CVS
 TK> repository
 TK> (module imlib) or from ftp.labs.redhat.com.  There is a tar file and an
 TK> rpm
 TK> somewhere under /pub/gnome/devel.  Don't get the one's from /pub/Imlib --
 TK> they're old and don't work.
i will try that. maybe someone should tell this to the guys who update the
gnome and/or rhlabs webpages?


 >> - gtk
 >>   you *need* 0.99.5 or later, even though gnome says 0.99.3 would
 >> suffice. it doesn't, i have tried.
 >>   also, in 0.99.9 (other versions unknown) there is a bug in the gtk
 >> install script. it will not install glibconfig.h, it simply forgets about 
 >> it. just copy glibconfig.h where glib.h went by hand.
 >> 
 TK> glibconfig.h actually does get installed -- but it gets installed
 TK> someplace
 TK> under $prefix/lib -- Why?  Well, the reasoning by the gtk+ maintainers is
 TK> glibconfig.h is system dependent, therefore it shouldn't go into
 TK> /usr/include
 TK> with the system independent headers....
makes sense, but otoh you get errors about it not being found that way.
shouldn't include files go to $whatever/include ? just for consistency?


 >> - mico
 >>   and mico *does* need egcs. it will compile with gcc, but you'll get
 >> problems later on with /very/ odd "unresolved references" (L155780 
 >> anyone?). 
 TK> I remember seeing this problem on the list before.  It's not a gcc/egcs
 TK> problem.  It's something to do with the command line options when
 TK> compiling
 TK> mico or not using mico-c++ (a script/wrapper around your c++ compiler) I
 TK> forget which.  Hopefully someone else on the list will clarify.
sure? i didn't change a single thing inbetween, only used egcs instead of gcc.
so whatever the problem is, switching to egcs seems to solve it.