On Tue, Jan 21, 2003 at 02:59:07PM +0100, devik wrote: > > Can you produce a "binary diff" of sent and received file to see how > > exactly they differ? Are the two files completely different? Or do > > they differ just in a couple of places? Is something missing? Is > > something changed? If so, how? That's how I managed to isolate some > > difficult data-dependent bugs, it might be useful here, too. > > I did (I wrote program for it). 16MB are totaly dirrerent > (from offset 14MB aligned on page boundary) and some other > part (10kB) is shifted by 1 byte. That part is not even 512B > aligned. This is list of different parts between 148005111 > long files f1 and f2: > from 14557184, cnt 4481067 > from 19038287, cnt 11393 > at 19038288 (by 1) in f1 > at 19038286 (by -1) in f2 > from 19049716, cnt 1162 > at 19049717 (by 1) in f1 > at 19049715 (by -1) in f2 > from 19050914, cnt 46210654 > from 68514564, cnt 4855808 > at 68518659 (by 4095) in f2 > > "at" block is where the block was found in other file. Uff. ;) I give up, I don't remember seeing anything like this. I would do one tcpdump at the sender and another at the receiver and compare the files. If they differ, something's changed along the path, if they don't, it should be possible to find out which version of the file the dumps' data correspond to. But that's what you've probably done already ... ;) pvl
