GitLab

Use clone customisation and test it on a real clone in the data.

system is in systemBox not nameBox

On our server the original_names in the .vidjil and .analysis files do
not match. Hence we must not claim that they must match!

Use real example (and explain it) for clustering.
It will be tested that clones are indeed clustered.

Make it consistent with the test in model_loader.js

How the server gets the files. It's not just a download of the stored
files.

The documentation says that the analysis overrides what is in the Vidjil
file. We do so.

Loading analysis was not tested. Hence some problems with original_names
or timestamps not replaced by the content provided in the analysis.

The example is changed so that we have an example of order change (that
will be tested)

Access to tag selection (to hide or show tags)

Function to access graph legend (x axis) in the Vidjil browser (for
functional tests)

Function to access external data in the Vidjil browser (for functional
tests)

The analysis was not properly loaded (we didn't use that yet for the
functional tests).

1. It is not very well supported on web browser:
http://caniuse.com/#feat=input-number
2. FF forces the decimal separator to
the one in the language of the user, which is non-sense on an english app
(where we expect the period to be the decimal separator). This may be
parameterable but I didn't find out.

Even if the Vidjil browser is quick we may not need to increment a
variable just for fun.

A temporary file is created in CGI mode to store the FASTA on which we compute
the similarity matrix

This is perhaps not the best solution, as should-to-tap.sh becomes specific to these tests.

This will be more flexible.