Andrius Merkys

Build failure with 'make --shuffle=reverse'

Thanks. Is it still possible to use OSRA with unpatched GOCR? Are the changes in the patched version substantial, or just concerning threading support?

Build failure with GOCR: thread-local declaration of 'job_t* OCR_JOB' follows non-thread-local declaration

I am using provean v1.1.5 without the supporting set and it always fails with the following (timestamps removed): [...] retrieving subject sequence information... [...] Parsing Error I have inspected the raw blast_out file and it seems fine: there are lines containing database IDs. Then I looked into the code in src/SequenceDB.cpp, in SequenceDB::SetSequenceInfoFromBlastOut(), and I do not quite get the following line: gi = strtok(NULL, "\t\n"); I may not know C/C++ well, but it seems to me that...

Thanks for the answer, Igor. I now see your point, and it makes much more sense to me.

Sure, this is an option. However, outdated configure.ac might at some point start to hinder the development of the configure script as all modern Autoconfs will refuse reading it. I am aware that configure script does not change much, though.

Yes, the original configure script works. However, Debian package builder attempts to automatically run configure with default Debian options, some of which are not supported by OSRA configure, in particular '--runstatedir=/run'. Rebuilding configure with Autoconf usually makes all such options available. I can in principle strip '--runstatedir=/run' off, but this would be something of a hack in Debian package build.

Autoconf directives need updating