- Event Series
- This event repeats every week.
- no-double-fork-needed, but function name change?: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=868550#45
- in gniibe/scd-kdf-support branch: T3152: KDF DO support in OpenPGP card
- will apply to 2.2 branch:
- D452: Build: FreeBSD make: Not for master, since it will require newer libgpg-error with yat2m
- D450: clock_gettime if CLOCK_THREAD_CPUTIME_ID is available.: Ready to 2.2
- T3351: libgcrypt: t_secmem fails on system with large page size (ppc64) by rC621f5c4e8373: Fix secmem test for machine with larger page.
- T2923: trust signature domain restrictions don't work by rGccf3ba92087e: g10: Fix regexp sanitization.: Should be in 1.4 and 2.2 too.
- This week:
- suspend/resume of USB -> support in Gnuk, experiment with scdaemon
- Goldilocks field computation: add, mul -> towards X448
re: double fork. Simply removing it is IMHO too brittle. We introduced it very early to avoid all kind of problems which hanging processes due to ill-behaving programs using gpgme. OTOH, there is still the problem on to reliable close file descriptors in multi-threaded programs. We now have a solution for Linux but it is Linux only.
On Windows we use a spawning helper which wraps the execs and closes the file descriptors - this is the only way to do this on older Windows versions Since Windows-7 there seems to be a system call to close file descriptor (HANDLEs) more selectively. That would allow to remove the spawning helper. OTOH, on Unix a spawning helper could be used to avoid the double for and would also simplify file descriptor closing.
Some people don't like extra wrapper programs - but in reality the overhead is very small.