GitLab

see #4275

benchmark-releases.py: exit with 42 when the last release/HEAD has significant difference with the previous one

closes #4282

We cannot use time on some commands such as the ones with cd

Use /usr/bin/time to get accurate memory resource usage. rusage couldn't be reliably used because
it was influenced by the parent process.

We take the minimal among the runs as the program is deterministic the minimal value is the closest to the execution time without external
perturbations.

Fix #4275

see #4278

Bug remained in some cases because of a poor initial test case

Fix #4272 (and #4263)

see #4268

If we try to set the same index several times we will wrongly have several references to it and, thus,
will never be deleted, which Valgrind dislikes

See !674

Reset the seed to have the same conditions. Otherwise depending on the number of tests and the number of times
the pseudo-random generator has been used the test may fail or succeed (we are using a RandomScore now)

We didn't take into account the length (as it was added later). This is now fixed.

see #4270

opens #4270

Our main file is .vijdil, and not anymore .vdj.fa (for ages)

see #4269

opens #4269

For some reason it was failing with my g++-5

As trimSequence is used on a representative and as the representative should contain the window we now
make sure that the window is not trimmed.

Fix #4263

thanks to @mikael-s