Hi, ok I think the bug's closed now by upstream's new version. Just uploaded it after some more tests than usually. David tells me, his images are read well now.
Thanks from Erik (upstream) to the valuable hints in this bug report and thanks to him for the fast fix. The problem was in jpegutils.c. > We could fix this one. It is indeed related - as mentioned in the patch mail > - to the static allocation of [ 1024 * ...] buffers just as the comment in > jpegutils.c says "accomodate up to 1024-wide images". ;-) > > A dynamic allocation seems prone to memory leaks. > > Perhaps we 'll just increase it to 4096 and try to catch images of bigger width. > > Greetings, > Erik If nobody finds something really bad in the code I'll ask (beg?) for inclusion in etch. Merry Xmas, Roland -- ------------------------------------------ Roland Marcus Rutschmann <[EMAIL PROTECTED]> ------------------------------------------
pgp7nPah8HHdc.pgp
Description: PGP signature
