Memory limitations

Almost all processing steps are available spectrum per spectrum because this is a very powerful way to work around the limitation of your computer RAM memory. There are two main exceptions:

• When opening a file, CLASS is buffering information (source name, line name, telescope name, scan number, offset, etc...) on each observations of the file to speed next find commands. This buffer has a fixed sized to avoid code complexity. This limits the number of observations that CLASS90 can read/write during one session. The default maximum number of observations is currently set to $100\,000$ which amounts to about 5 MB of RAM memory. This number can be set to a larger value through the CLASS_IDX_SIZE logical variable in your $HOME/.gag.dico configuration file as it is probable that the default value is too small with the largest maps observed today.
• In CLASS90, it is also possible to load a whole index as a single 2D array for further visualization and processing (see below). There is no limitation on the number of dumped spectra loadable, apart from the previous limitation and the RAM memory of your computer.

Except from the index buffer, all other CLASS90 buffers are now dynamically allocated, in particular the buffer R & T containing the spectra data. This means that the number of channels of a spectrum is now unlimited which is a useful feature for line surveys.