Session introspection and managementΒΆ
The sessions created by pymanip use standard formats for storage:
the synchronous sessions use both HDF5 and raw ascii files. For a session named toto, three files are created: toto.dat, toto.hdf5 and toto.log;
the asynchronous sessions use a SQLite3 database. For a session name toto, one file toto.db is created.
The rationale for having three files in the first case is that HDF5 is not designed for repeated access, and it happens sometimes that the file gets corrupted, for example if the program is interrupted while the program was writting to the disk. The ascii files however are opened in append mode, so that no data can ever be lost.
We switched to a database format for the newer asynchronous session, because such a format is designed for multiple concurrent access, and each transaction is atomic and can be safely rolled back if an unexpected error occurs. The risk of corruption is much less, and we thought it was no longer necessary to keep the ascii files.
Because all those file formats are simple and documented, it is possible to read and inspect them with standard tools, such as h5dump for the synchronous session files, or the sqlite3 command line tool, but it is impractical to use these tools to quickly check the content of some session file. That is why a simple command line tool is provided.
The main command for introspecting saved session is
$ python -m pymanip info session_name
It can read asynchronous sessions, as well as synchronous sessions and old-style OctMI sessions. It is not necessary to happen the filename extensions, e.g. .db for asynchronous sessions. The command will output the start and end dates of the session, as well as a list of saved variables.
Two other commands are provided, but they are specific to synchronous sessions which are stored in HDF5 file format: