GitLab

See #2881

due to issues with SSL certificates and Watir, use HTTP.

Adds a defs.py variable to indicate wether https is required for server
requests

See #2881

we don't use it.

Allows the docker container to be built from the git repository rather
than copying files. The desired branch and repo address can be passed as
build parameters to overwrite the default values

See #2940

See #2881

Adds a ci job for building and testing a docker image and a job for
deploying a docker image
See #2881

instead of using the debian packages simply copy the vidjil source
code into the container. This way it is easier to automate a docker
image build.
See #2940

Display the number of k-mer for each genes if debug_filter is defined

Closes #3346

See merge request !236

For more informations see #3346.

See #3259.

Reverts commit c0d14efb.
We don't need this extreme solution: coverage can be quick if not computing
branch coverage.

Coverage is now much quicker. The additional time for coverage over all the
jobs is about 3 min.

Fixes #3339

Perversely chosen mutations may fool our best heuristics.
See #3297 and #3298.

.gitlab-ci.yml: move coverage analysis to the end, set it manual

See merge request !233

See #3339. We want to have first fast tests.
Once we will have #3340, we will run again always this stage.

Feature a/3259 output filtered bioreader informations

Closes #3259

See merge request !231

See #3259.

Since we display the filter's informations in the main, it makes more sense
that this class has its own display method.

For more informations see #3259.

Since the original BioReader is called through every Filter's functions, it made
more sense to add it as an attribute in its class.

For more informations see #3259.

The Filter's informations are displayed for each germline at the end of execution.

For more informations, see #3259.

For more informations see #3259.

Both of them are attributes stored in the Filter class.
filtered_sequences_calls is the number of times filterBioReaderWithACAutomaton is called
and filtered_sequences_calls is the number of sequences returned by this method.

For more informations see #3259.

Merge branch 'feature-a/3282-repetitive-filterbioreaderwithacautomaton-parts-to-reusable-function' into 'dev'

Feature a/3282 repetitive filterbioreaderwithacautomaton parts to reusable function

Closes #3282, #3312, and #3299

See merge request !228

The test verifies that when an odd K-mer is transmitted (unknown, ambiguous, wrong ascii n° for its label...)
an exception is thrown. Also it verifies that when a K-mer is transmitted, there is the good amount of sequences
that are transfered from a BioReader to another.

For more informations see #3282.

Since the Filter object is unique for a Germline, there is reason to delete it many times
in a loop. Also to simplify the test, it has been completely removed.

For more informations see #3282.

For more informations see #3282.

Factorize filterBioReaderWithACAutomaton with transferBioReaderSequences and use isGeneric/SPECIFIC_KMERS_NUMBER

For more informations see #3282.

This method is used to transfer sequences from a BioReader to another based on
a KmerAffect.

For more informations see #3282.

isGeneric is a method returning true if the current K-mer is not odd (ambiguous, unknown).
It is used instead of calling successively isAmbiguous() || isUnknown()...

SPECIFIC_KMERS_NUMBER is a constant representing the amount of odd K-mers. For now there is
only two (ambiguous/unknown) but in the future it might grow.

For more informations about it, see #3282.

Coverage reports

Closes #1696

See merge request !227

They are thus more easily accessible through Gitlab interface

Depending on the activation of Gitlab Pages (and on my
own understanding of this feature…)

This is better than specifying it in the project settings.

As pointed out by @magiraud the files may not be present
anymore on the slave to analyse the coverage. We therefore
need to pass them as artifacts.

Tests are now launched with coverage included.
A specific job will recover the coverage

lcov needs to know the whole path to a file in order to display it
and to count the coverage on each file.

However when launching with Gitlab-CI, each job is launched in a separate directory
which leads to different paths for the same source files.
Therefore we first remove the trailing path when generating the
lcov files. But we add it again before generating the HTML output
as it requires to access the files.

This prevents from re-using old data.