Sandbox repo for trying out various things.
The gh-pages
site is deployed at this address, although might not contain anything as the website build has been delegated to another repository.
The "profiling" workflow is triggered on pushes to main and on request.
This workflow calls the benchmark.py
script which produces .pyisession
files containing the profiling outputs alongside some metadata files.
These files are pushed to the target
branch of the willGraham01/benchmark-action-playtest-target
repository, and a repository_dispatch
event is sent to trigger the website build in that repository.
Output files from profiling have the same filenames but different extensions to distinguish them;
- The
pyisession
file contains thepyinstrument
profiling output. - The
.json
file is contains thejson
output of the profiling session. This can be obtained from theipysession
file, and so does not need to be pushed. - A
.stats.json
file containing metadata that we otherwise want to pass across to the website build, but which cannot be obtained from the.pyisession
file. These are things like the commit SHA/hash, the workflow event trigger, pull request target, etc.
To work, this repository need a PAT
token available as an actions
secret.
Said PAT
token needs at least read/write access to the contents
of the target repository, as well as metadata
read permissions (as otherwise the website rebuild cannot be triggered).