GitLab

as in algo/Makefile

Contains sequences from other files (redundant)

Those files contain fake sequences.

This merge all .should-vdj.fa files into one file, and test this file.

This test is twice faster than an usual 'make shouldvdj_with_rc': it should
be used to quickly check whether these tests run. However, when something fails,
it does not provide precisely the location of the error.

Now all should-vdj tests are also run on the rev-comped sequences.
This increases the test time, but improves our robustness.

When entering mnaually curated sequences, this is faster than launching
a full 'make shouldlocus_and_vdj' (which further recompiles everything).

"$(CXX) -M > dep.mk" also needs a "-std=c++11" flag

Since cb9d3fea, shouldvdj errors were not taken into account.
This is fixed.

We should not launch a command directly but use variables instead.
The standard variable for make is $(MAKE). It is therefore more portable
(for instance on freebsd where the make command used should be gmake)

All KmerSegmenter should-vdj tests (-2q) are now also run on revcomp'd files (-r), and
should give exactly the same output than the original test files.

Even if it works, we choose to not run FineSegmenter tests with the same option.
It would test nothing more, as the '-c segment' command first calls the KmerSegmenter to select
the best locus with its strand.

cleancoverage removes coverage-related files
cleantests removes test-related files
cleanall removes everything.

We want to be able to test coverage on different rules, without
deleting previous coverage files.

In case of spaces and parentheses, the path must be protected for mktemp but also
for the --xml-file option. Hence the double double quotes.

The outer double quotes must not be escaped (otherwise double quotes are passed to valgrind
which is not good)

The dir function was considering spaces as separator, which is bad.

Makefile: Allow to launch a single functional test with valgrind by prefixing its name with valgrind_

This has no consequence since this rule is forced anyway.

Relative path is less robust when changing directory.

valgrind_should: valgrind on functional tests
valgrind_tests:  valgrind on unit tests

This invocation should also be more portable
See http://stackoverflow.com/questions/2792675/how-portable-is-mktemp1

Makefile: For tests ensure that vidjil is rebuilt with -O0 -g so that we get useful information with Valgrind (slower...)