#4003: Look further back in time per op file to see if we had a prior…

[eyonland]

… failure

[tt-rkim]

Hey @eyonland
@warthog9 and I discussed what possible security issues could be going on here.

The conclusion is essentially that yes, while someone could put something nefarious into any build scripts, this would apply for anyone who has access to our codebase. Nothing specific about this build process should disrupt us.

I'll help with integrating into the docs build from here, and show you what it looks like after.

[tt-rkim]

• Move the ttnn_sweeps directive before the actual sphinx make docs step in the GH workflow build
• Get rid of the checked in sweeps rst and change ttnn_sweeps/index.rst to a placeholder file
• Request a read token for GITHUB_TOKEN (lmk)
• Ensure that generated rsts are properly formatted so build doesn't break
• Ensure that people can still build locally with the ttnn_sweeps, but that the pipeline build correctly builds with ttnn_sweeps
• You can verify the whole flow on CI by downloading the gh_pages artifact on branch and serving it to yourself on localhost with python -m http.server <PORT> in the unzipped build folder

cc: @vtangTT

@eyonland - @vtangTT will try looking at it today, but he is also working on priority cmake work. We will reconvene on infra team later today to look at progress. If this is urgent P0, please loop in Jason

[tt-rkim]

Why did you close this? We're actively working on it. @davorchap
Please read the actions more carefully next time to check if you should close.
I noticed you did this with a lot of other PRs. Thanks for doing that.