Log filter v2 #188
Log filter v2 #188
Conversation
|
Nice! I think the water statements are a bit too much too. Perhaps reformulating them to something more general about the route? |
VeckoTheGecko
left a comment
VeckoTheGecko
left a comment
There was a problem hiding this comment.
Have you considered using https://github.com/pavdmyt/yaspin ? Looks like a good package with options for this. It has minimal dependencies and is quite well maintained.
| # try/finally to ensure filter is always removed even if .execute fails (to avoid filter being appled universally) | ||
| # also suits starting and ending the rotator for custom log message | ||
| try: | ||
| rotator.start() | ||
|
|
||
| for point in sample_points: | ||
| particleset.lon_nextloop[:] = point.location.lon | ||
| particleset.lat_nextloop[:] = point.location.lat | ||
| particleset.time_nextloop[:] = fieldset.time_origin.reltime( | ||
| np.datetime64(point.time) | ||
| ) | ||
|
|
||
| # perform one step using the particleset | ||
| # dt and runtime are set so exactly one step is made. | ||
| particleset.execute( | ||
| [_sample_velocity], | ||
| dt=1, | ||
| runtime=1, | ||
| verbose_progress=False, | ||
| output_file=out_file, | ||
| ) | ||
|
|
||
| finally: | ||
| rotator.stop() | ||
| for handler in external_logger.handlers: | ||
| handler.removeFilter(handler.filters[0]) |
There was a problem hiding this comment.
In Python, this try finally pattern is normally done via context managers. This means that if something fails in the block of code, the code is torn down after.
This will simplify both the code for the spinner, as well as the code for the filtering, boiling down to something like
with spinner("Deploying drifters..."):
with discard_parcels_warnings():
# code to process measurement
This PR is an updated version of #185 (now closed), originally relating to #155. Rather than simply removing duplicate messages in the log, all Parcels messages are now suppressed to decouple from Parcels, as suggested by @VeckoTheGecko in #185.
I have made further updates to what's printed for the user as they run VirtualShip. This includes adding spinning dials to most instruments to provide a visual cue that the measurement simulation is running (point 2 in #186). For the instruments which take longer to simulate (drifters and argo floats) I have left the call to use Parcels
verbose_progress = Trueto get a progress bar. I would have liked to have decoupled from Parcels again here but it proved tricky to propagate our own progress bar given the iteration occurs within the.execute()and it is difficult to access this or replicate the iteration outside of the Parcels internals.Finally, I've reorganised some of the print statements in do_expedition.py to provide a bit more structure to what's printed for the user. It's subjective but hopefully it improves readability and tracking the simulation progress a bit? See an example screenshot below for a completed expedition (the spinning dials are replaced with "[COMPLETED]" when the execution is finished).
The Parcels progress bar doesn't fit in so nicely but maybe something for a future PR...