pandalyzer

[matthiasdiener]

Here is how it looks like:

$ pandalyzer foo.sqlite
Pandalyzer running on Python3.8.6 | packaged by conda-forge | (default, Oct  7 2020, 18:42:56)
Run help() to see help for 'magic' commands.
>>> help()

Commands:
 help()       show this help message
 runprops()   show a list of run properties
 quantities() show a list of time-dependent quantities
 warnings()   show a list of warnings
 dump()       show an SQL table

Plotting:
 dbplot()     plot list of quantities.

Available Python symbols:
    db        the database

>>> runprops()
Property       | Value
---------------+----------------------------------
id             | 1
dirname        |
filename       | mpi-log.dat-rank0-TRBefX
schema_version | 2
is_parallel    | 1
unique_run_id  | cb5b07f831d311ebb4ba784f435f24cb
rank_count     | 4
myconst        | 0.6468262878955677
cmdline        | log-mpi.py
machine        | MDmbp
date           | Sat, 28 Nov 2020 17:45:28 CST
unixtime       | 1606607128.460344
>>> quantities()
id | name    | unit | description                 | rank_aggregator
---+---------+------+-----------------------------+-----------------
1  | t_step  | s    | Time step duration          | None
2  | t_2step | s    | Step-to-step duration       | None
3  | t_cpu   | s    | Wall time                   | None
4  | t_log   | s    | Time spent updating the log | None
5  | step    | 1    | Timesteps                   | None
6  | t_sim   | s    | Simulation Time             | None
7  | dt      | s    | Simulation Timestep         | None
8  | t_vis   | s    | Time spent visualizing      | None
>>> dbplot(["step", "t_vis", "t_step"], kind="line")
<AxesSubplot:xlabel='step [1]', ylabel='t_vis [s], t_step [s]'>
>>>
now exiting PandalyzerConsole...

[inducer]

It's not obvious to me why this functionality needs to be separate from the existing tooling. If branding is important to you, then sure, kill runalyzer, and let this survive.

[matthiasdiener]

It's not obvious to me why this functionality needs to be separate from the existing tooling. If branding is important to you, then sure, kill runalyzer, and let this survive.

This PR is just me playing around with the pandas functionality to see if it might be useful. Imho, even if we decide to merge this at one point, keeping both runalyzer and this around could be useful (e.g., for people used to the runalyzer API).