Juliusz Chroboczek wrote:
>
> I have recently tried out GNUstep 0.6.5 with the xdps backend and the
> free DPS extension available from
>
> http://dps.sourceforge.net
>
> After fixing a bug in the DPS extension and forcibly disabling the
> transparency support in the xdps backend, clients would start, go as
> far as drawing the main menu, and then silently quit. The PostScript
> trace shows that the clients convert a point from user space to device
> space, then invokes ``grestore'' four times in a row, and then
> silently quits. No crash, no error message, no debug log.
>
> Any hints with debugging this are welcome.
>
> While I'm at it, I'd like to report the following xdps bugs:
>
> * detection of the presence of the NeXT extensions is unreliable;
> * determining the list of fonts is not portable to the Adobe
> implementation;
> * determining the list of fonts is inefficient (it causes one
> round-trip per font);
> * conversion between userspace and device space is inefficient; the
> CTM should be cached in the client, and no round-trips should be
> necessary;
> * the xdps backend uses DPSPrintf in quite a few places; wraps should
> be used instead.
>
I had started fixing some of that, but got distracted. I don't know much
about font stuff, so if you can give me any help there, that would be
nice. In particular, I think that xdps uses a different mechanism if it
detects Adobe, but it would be nice to use the same one for anything.
Not sure about the crashing, but I'll try to see what I can find.
--
Adam Fedor, Digital Optics | Fudd's law of opposition: Push
[EMAIL PROTECTED] http://www.doc.com | something hard enough, and it
[EMAIL PROTECTED] http://www.gnustep.org | will fall over.
