> The try throws an error that the catch doesn't catch (not sure what
> yet; it's hard to track things cross-language). I'll post a patch once
> we get this figured out.
Update, for anyone who might be interested:
This seems to be a straight invalid memory reference coming from
JpegBase::readMetadata(). We're trying to figure out where it's coming
from.
My project, a .NET 2.0 app that uses exiv2 (via exivsimple), occasionally leads to exiv2 causing an AccessViolationException. Before we moved our app to 2.0,...
... [...] ... ...might well be, this is a common problem in C and C++ programs. Unfortunately, they are difficult to find. To detect such bugs, I use Valgrind...
... don't cover ... Actually (amusingly enough), I'm the reporter on that ticket. We're using a sligthly customized version of the library that fixes this ...
Ok, so it looks like this bug pops up *more* often when we are using exiv2 to do read lots of pictures "at once". We are locking access to the library, due to...
... Ok, thanks. So I think this is due to exivsimple's lax file-access checking (we probably should have made it more robust before using it directly). If...
... Update, for anyone who might be interested: This seems to be a straight invalid memory reference coming from JpegBase::readMetadata(). We're trying to...
... I don't think so. At least three of my team members have been able to consistently reproduce this on totally different picture sets. I haven't tracked my...
... I've checked, and it happens on various images. I've tracked it down to this: int scanned = sscanf(buf, format, &t.hour, &t.minute, &t.second, &plusMinus,...
Another update: a member of my team tried ripping out all of the relevant exiv-calling code and put it in a absolutely-definitely-single-threaded driver class....
... Is this code something you could send me to along with the images? I would not keep the code or the images around after looking into this problem. Also did...
... happen? ... I should be able to, but not before this weekend at the earliest. Actually, the stripped .NET exposes some interesting determinism underlying...
... Can you share these two pictures? I'd like to have a closer look at them. You could upload them to the Files or Photos section here or send them to me...
... BasicIo::cur); ... don't fully ... Nope. Hits the same error, unfortunately. I'll start on the C++ app to reproduce this tomorrow night--hopefully that...
... Ok, next try... Valgrind reports a problem at the same location that you found, within that sscanf call. This patch assumes sscanf reads beyond the end of...
Well I'll be a monkey's uncle. I think that's it. The test scanner does not crash anymore (and it used to crash consistenly on those two pics), and our app has...
... Ok scratch that, I remembered after sending that these methods are used to read data out of the middle of buffers that won't have nulls in them. Maybe the...