Open darkquasar opened 4 years ago
@darkquasar I already updated the readme file, I will leave the update of the index file in your hands. Please add a description of the visualization of BestiaPop running, I have added a new section called BestiaPop in action! for this purpose.
I will create a gif of BestiaPop running, that always catches the eye.
Regarding the performance tests, it would seem that multiprocessing is always slower than the alternative, which seems kind of odd, and would render multiprocessing kind of useless. Can I ask you to run the tests but instead of testing with 36 lat/lon combinations to increase that to at least 100 combinations and at least 5 years instead of 1. So this 36 lat-lon combinations (36 files) for a single year (36f1y)
would be 100 lat-lon combinations (100 files) for 5 years (100f5y)
. This is because of the way that BestiaPop fetches data from the cloud APIs, multiprocessing should be more effective the more lat/lon combinations you request since each combination is a call to cloud APIs.
@JJguri it would be great if you could also have a go at updating index.rst. I can jump in after you've tried to add the required content and help if things don't go well. Here are some pointers:
index.rst
is written in ReStructuredText, it's just another markup language, required by ReadTheDocs and to auto-document python modules. It's simple really, here is a cheatsheet that you can go for references.index.rst
and commit the changes, wait 5min and the changes should reflect in the ReadTheDocs site. That's one way of finding out if it looks how it's supposed to look.pandoc
to convert online from current markdown (.md) content to reStructuredText (.rst) content, see here. Note: only do this for the sections that you want to add to index.rst, avoid copy/pasting the whole readme.md since it might create issues. Instead of using #
for H1, use=
underlining the title. The quantity of =
symbols needs to be as long or more than the length of the title being underlined. Example:
This is my Title
================
Instead of using ##
for H2, use ~
underlining the title.
Instead of using ###
for H3, use +
underlining the title.
For links, instead of doing this [my link](https://mylink.com)
you need to do this
`My Link <https://mylink.com>`_
I will create a gif of BestiaPop running, that always catches the eye.
Regarding the performance tests, it would seem that multiprocessing is always slower than the alternative, which seems kind of odd, and would render multiprocessing kind of useless. Can I ask you to run the tests but instead of testing with 36 lat/lon combinations to increase that to at least 100 combinations and at least 5 years instead of 1. So this
36 lat-lon combinations (36 files) for a single year (36f1y)
would be100 lat-lon combinations (100 files) for 5 years (100f5y)
. This is because of the way that BestiaPop fetches data from the cloud APIs, multiprocessing should be more effective the more lat/lon combinations you request since each combination is a call to cloud APIs.
@darkquasar please have a look at the updated table for performance in the documentation. You were right, the use of -m
is recommended when the lat-long combinations are high.
Hey @JJguri, I've re-structured the documentation in README and index.rst to reflect the new functionalities of the package. There are some sections though that need further touches or information to be expanded. I've flagged these with
[yoni to expand]
comments in the files.Please check and action as required, if no further info is needed, then delete the comments. It would be good if you could also make sure there is cohesion in the text.