Kamil Dudka wrote: > > The suggested patch is attached, > > Bad news. The patch breaks it completely. > > Have you tested it yourself first?
No. I did some thorough testing now and found a stupid mistake that aused the 100% failures. But then it starts to fail intermittently, and with some added debug prints I found out that my idea just doesn't work, even though it seems like the right thing to do. gcry_sexp_nth_mpi() simply doesn't understand the data that is returned by the sign function. The #gnupg channel was sufficiently idle when I looked there that I just give up on this. Other parts of the code are using the same construct of directly parsing the signature sexp after calling gcry_sexp_nth_data() and I don't want to spend more time now on fixing something that works. :) Thanks for testing and helping fix the bug! //Peter _______________________________________________ libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
