pfdInitConverter

New Message Reply Date view Thread view Subject view Author view

From: Dirk Scheffter (scheff++at++iff.fhg.de)
Date: 08/24/2001 08:58:54


pfHi,

we have a Performer application in which the user can assemble VR worlds
interactively.

However, because we cannot foresee which files the user will load I have
implemented a function which looks into all relevant directories and
extracts filenames which matches the name scheme of a pfConverter lib.
(It uses a search routine like described in man pages of
pfdInitConverter.)

If a file name matches the name scheme the supported extension is
extracted from this filename and used with pfdInitConverter.

Next problem is to find out what abilities has the converter (for
example if it's able to store)
Because I found no other (better) solution, I added a function to the
pfdLoadFile.c source which looks into the internal table with entries of
type ConverterDSO to tell what functions are available.

To be sure, that the customer will use the right libpfdu (and not force
him to botch around in the system) I linked my extended libpfdu,
statically.

However, it worked fine for months on IRIX. On Linux I cannot get it to
work! I have read a lot about dlopen and such stuff, have increased the
PFNFYLEVEL upto 7, but nothing helps.

On Linux the converter seems to be found (dlopen() returns handle) but if
function pointers are requested (dlsym()) none valid pointer is returned.
pfdInitConverter() allways fails.

If I use libpfdu.so which was delivered with Pf2.4 (linux) then the
converters do their job like expected.

I have no more ideas.

How can I
(A) compile sources of libpfdu on Linux to get it working?
(B) request abilities of converters using original libpfdu?

Every hint is very, very appreciated.

(I could even accept the promise that SGI will provide the above described
feature in the next release of Performer. This is combined with
the hope that this will happen before CeBIT 2002...)

Sorry for the long story, but I'm really frustrated about stumbling around
through boring system stuff instead of doing my real work... ;-(

Best regards and a nice weekend,
Dirk.

P.S.: Sorry Brian, I really tried it...

-----------------------------------------------------------
Dipl.-Inf. Dirk Scheffter scheff++at++iff.fhg.de
Fraunhofer IFF fon: +49 (0) 391/40 90-854
Sandtorstr. 22 fax: +49 (0) 391/40 90-870
D-39106 Magdeburg http://www2.iff.fhg.de/~scheff
Germany
-----------------------------------------------------------


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Fri Aug 24 2001 - 08:58:32 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.