I've got an image which will cause gocr-0.40-r2 to segfault along with a patch which fixes the issue. I'm not sure if this is related to bug 145624 or bug 153068 but the patch fixes the problem with this image. I'll attach the image/patch/straces.
Created attachment 103354 [details] Image which causes segfault
Created attachment 103355 [details, diff] Patch which fixes it
Created attachment 103357 [details] strace before applying patch
Created attachment 103358 [details] strace after applying patch
This patch has come from the FuzzyOCR guys and is one of the things holding up bug 154392, although I'll probably add the FuzzyOCR ebuild with a warning that some images may cause a segfault until this bug is fixed.
I've just checked gocr 0.42 and this problem has been fixed (in a slightly different way) so having an ebuild for the 0.42 would also fix the problem. If I get some time I could put one together unless some who knows the codebase sorts one out before I do.
(In reply to comment #6) > I've just checked gocr 0.42 and this problem has been fixed (in a slightly > different way) so having an ebuild for the 0.42 would also fix the problem. If > I get some time I could put one together unless some who knows the codebase > sorts one out before I do. > Marking as fixed as 0.43 has been added to the tree as per bug 145624.
