glob is imported in the pcigale/session/configuration.py to allow its use in
the configuration file (with eval).

TODO: write a new configuration pattern to add several files with something
like "files *.fits".

There is maybe a bug triggering the display of the best model figure in
a matplotlib interactive window.

Use a simplified gaussian kernel density estimation for computing the
probability density functions.

The team will use the projets.lam.fr issue tracker instead.

Modification to savefluxes to have all the contenct of the sed.info
dictionary in the output table. That means that all the SED created with
the modules must have the same sed.info keys.

For a given observation, taking the least chi-square or the least
reduced chi-square is the same thing.

Start of the rewrite of the psum analysis module in pdf_analysis. The
computation is made avoiding loops by using numpy multidimensional arrays.

Use np.max/np.min instead of the built-in functions. This results in a factor ~2 speedup to compute the fluxes.

Maybe we should change the name.

To make it easy to save the best models table.

Factor 2 speedup to compute the IGM attenuation. This is done by doing some computations only on the part of the SED actually affected by the IGM. This could still be improved.

Add redshifting and IGM attenuation modules (make psum use them).
Remove atpy dependency and use astropy instead.
Some modification to make the code Python 3 compatible.

Correction of a problem spotted by Médéric, basestring does not exists
in Python 3.

Make the source code more compatible with Python 3. As far as I understand the documentation, this should be harmless for Python 2.

and remove old igmattenuation code.

Implementation of one module for the redshifting and one for the IGM
attenuation. For now, the IGM module only implements the Meskin (2006)
attenuation; alternative attenuation law can be added with a parameter
in the module to choose which one to apply.

Extend the Calzetti-Leitherer with the Leitherer formula below 91.2 nm.
Maybe we should made the attenuation even stronger there.

Stop cutting the Maraston 2005 SSP below 91.2 nm.

Move igmattenuation module and external dependencies inside the psum analysis
module.

Remove redshift configuration from main configuration file.

Move the igmattenuation module and its LSST dependencies inside the psum
analysis module.
Remove the redshift_module and redshift_configuration parts of the
configuration file.

Put some function that may be shared between several analysis modules at
the analysis_modules level.

Making glob() available in session/configuration.py allows users of
savefluxes and sfhfromfile to use something like

filename = "eval glob('sfh*.fits')"
sfr_column = 2

to take the star formation history from several files.

Create two custom exceptions.

When one tries to insert something in the database that is already
inside, a DatabaseInserError is raised.

When one asks for something in the database that is not present, a
DatabaseLookupError is raised indicating what was not found.