On Sat, 20 Oct 2007 17:07:14 +0000, walt wrote: > On Fri, 19 Oct 2007 23:39:25 -0400, Dave Chand wrote: > >> On Oct 6, 2007, at 6:58 PM, Greg Lee wrote: >> >>> I looked at some other 1 part images in this same newsgroup and found >>> that all those with byte size <= 369052 displayed correctly, but all >>> those with byte size >= 369361 displayed incorrectly... > >> I have been experimenting with gdk-pixbuf. What I find is the >> following: in the file gtk+2.12.x/gdk-pixbuf/io-jpeg.c there is the >> statement #define JPEG_PROG_BUF_SIZE 65536 >> if you change its value to something significantly higher, such as >> #JPEG_PROG_BUF_SIZE 500000 >> then it seems the decoding problems go away. > > I just sent this question to Matthias Clasen, who increased that number > from 4096 back in 2002. It seems that a pixbuf sets an arbitrary size > limit on how much data can be loaded in a single chunk. The real > question is whether that limit is intentional or merely a bug. > > Meanwhile, Brian Downing posted a patch to the pan bugreport: > http://bugzilla.gnome.org/attachment.cgi?id=96917&action=view > > His patch explicitly limits the amount of data Pan sends in one chunk, > and this is the correct fix at least until we know if this size problem > is a gdk bug or not.
I tried viewing an 8.5M jpeg using Dave's gtk+ patch and it did not display correctly. Also, I increased JPEG_PROG_BUF_SIZE further to 1000000, but it still didn't work right. Then I tried Brian Downing's patch on the SVN 324 version of pan2, and that worked fine for the large jpeg. I don't think this is going to surprise anyone, but I thought you might like confirmation. -- Greg _______________________________________________ Pan-users mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/pan-users
