Re: Changes to graphic context handling

[Mark Tracy]

Fred,
I tried the new svn 24903 with your changes. It looks like printing  
works now, when using the art backend. Cairo does not generate good  
EPS, but at least it tries.

On the related subject of Cairo, I just submitted 3 bug reports,  
including the printing problem. I submitted the TestApp that I used  
for debugging NSBitmapImageRep that you might find useful.

Cheers,
Mark Tracy

On Mar 19, 2007, at 6:29 AM, Fred Kiefer wrote:

> Mark Tracy wrote:
> > The change you made has broken one of my programs. I am printing  
> > from a
> > windowless custom view because I want to lay out the report for paper
> > not the screen. Maybe not a wise strategy, but certainly not an  
> > illegal
> > one. This works fine for Cocoa and for previous editions of GNUstep.
> >
> > In  svn 24892, NSView.m line 1900, there is:
> > - (void) lockFocusInRect: (NSRect)rect
> > {
> >     [self _lockFocusInContext: [_window graphicsContext] inRrect:  
> > rect]
> > }
> >
> > In my program, since the view is not associated with a window, nil is
> > passed, and I get a segfault trying to read the context.
> >
> > I changed it to read:
> > - (void) lockFocusInRect: (NSRect)rect
> > {
> >     NSGraphicsContext *ctxt =
> >       (_window ? [_window graphicsContext] : GSCurrentContext() );
> >     [self _lockFocusInContext: ctxt inRect: rect];
> > }
> >
> > This appears to solve my problem, but it should be reviewed by  
> > someone
> > who has a deeper knowledge.
>
> Hi Mark,
>
> thank you very much for this bug report. It shows that somebody is  
> using
> the print functionality in GNUstep. What you are doing is quite
> legitimate and GNUstep has to support it.
>
> I have tried to correct the problem you found in a slightly different
> way. Could you please have a go with current SVN source and report  
> back,
> if it still fails? There will be a couple of further changes to the
> graphics context handling in the next weeks. I try to do this big  
> change
> rather smoothly, but you already discovered the first error here. So
> please test this from time to time for new errors.
>
> Cheers,
> Fred