🚀 go-pugleaf

f80204@cad.ee.ntu.edu.tw (Hung-Chi Chu) writes:

>Hello,
>  I tried to compile iv-3.1 under Sun SPARCstation SUNOS-4.1.2 with X11R5 and gcc-2.3.2
>and everything seemed to be ok, but when I run doc it core dump.
>  Can anyone tell me what's wrong?  Please mail to f80204@cad.ee.ntu.edu.tw
>Thanks in advance.

>Chu.

The solution for me was to make resource a virtual base class for
glyph and handler. Mark Linton dislikes this for performance reasons,
but I don't know a better way to get it working with gcc. Of course
I'd also like to omit any performance drawbacks :-).

Thomas

> f80204@cad.ee.ntu.edu.tw (Hung-Chi Chu) writes:

>>  I tried to compile iv-3.1 under Sun SPARCstation SUNOS-4.1.2 with
>> X11R5 and gcc-2.3.2 and everything seemed to be ok, but when I run
>> doc it core dump.  Can anyone tell me what's wrong?

To which in article <tw.724961702@fgb1>, Thomas Wuensche <tw@fgb1.fgb.mw.tu-muenchen.de> replies:

> The solution for me was to make resource a virtual base class for
> glyph and handler.

There is a patch from Johan Garpendahl <garp@isy.liu.se> available at
ugle.unit.no:/pub/X11/contrib/InterViews/3.1-beta3-patch (I am sure it
can be found at Johan's "native" ftp server as well, isy.liu.se, but I
haven't checked). This fixes two things:
 o the Doc coredump 
 o the pattern menu of idraw (gives you the full number of built-in
   patterns. 
Read the instructions at the start of the file.

> Mark Linton dislikes this for performance reasons,

Well, I would have thought that virtual inheritance should give you
more efficient code memoryusage-wise. Obviously experience did show
that it wasn't so. I wonder why.

> but I don't know a better way to get it working with gcc.
> Of course I'd also like to omit any performance drawbacks :-).

I have been told by a trustworthy source that the virtual function
pointer table bug in g++ is being worked on. 

BTW, Merry X-mas everybody!

- Steinar