[ET Trac] [Einstein Toolkit] #1890: formaline capture simfactory information

[Einstein Toolkit]

#1890: formaline capture simfactory information
---------------------------------------------+------------------------------
  Reporter:  jonah.maxwell.miller@…          |       Owner:  eschnett           
      Type:  enhancement                     |      Status:  new                
  Priority:  unset                           |   Milestone:                     
 Component:  Other                           |     Version:  development version
Resolution:                                  |    Keywords:  formaline          
---------------------------------------------+------------------------------

Comment (by eschnett):

 Formaline is not just about storing the source code. Formaline is about
 making simulations reproducible, and that includes capturing all
 parameters (and parameter changes) at run time, and recording the
 simulation environment (machine, user, directory, time, various UUIDs,
 etc.), and also putting part of that information automatically into
 various places (ASCII and HDF5 output, unless disabled).

 I don't know whether Formaline records the number of MPI processes
 (probably yes) and the number of OpenMP threads (probably no), but it
 should record both, as well as the job id from the queuing system. It
 would be trivial to record the MDB entry that Simfactory used to create a
 simulation, and thus it should be done. "Recording" doesn't mean that it
 should be in one of the directories -- "recording" means that Formaline
 puts this information into all the output channels where recorded
 information is available. These are currently (mostly) a human-readable
 text file and a json file, as well as a database server if one is
 configured (it typically isn't, unfortunately). ((We should revisit this
 -- people find it quite easy to set up a private git repository, so we
 should go this route if it makes people use it.)) This information is also
 available at run time, if there is e.g. a web server or a log server
 running as a thorn. It should probably also go, in its entirety, into
 large output files.

-- 
Ticket URL: <https://trac.einsteintoolkit.org/ticket/1890#comment:8>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit