-
Notifications
You must be signed in to change notification settings - Fork 75
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Publish the history of the reports somewhere #771
Comments
@hzeller Any thoughts on what we should do here? |
I'd actually like to run all the tools in the different versions against the current set of tests. The tests were improved over time, have had fixes where there were plain wrong or tested the wrong thing, and also we extended the set of test-cases with new cores etc. So to have a nice apples-to-apples comparison, best would be to have the same set of tests (current state) run against all the historic versions of tools we have collected (I suppose we can get them from conda ?). While the single table that we have now is a good overview of a particular run, for the historic view, a couple of different views are also useful, such as a graph that shows tool-timestamp on the X-axis and passing tests on the y-axis. This would show how tools improve over time. Immediate tasks would be to change the CSV format to have a one row per (tool, test) tuple instead of one row per test with columns representing the tools. |
Seeing tool versions against head would be interesting. Note my suggestion was to simply keep all old reports because they have been (ideally) analyzed by "both sides" - tools and tests. Running against "head tests" would not solve the original reason for the request which is problems int he other direction - there used to be passes against Verilator (#770) and Icarus (#732) that are broken due to test driver changes. FWIW #446 is perhaps the best way (but more effort) to reduce test/driver problems. Though even if that is fixed I still think a simple history of reports has good value. |
So to see the reports we recorded, we could go through the history of the gh-pages:
... then So sending this to a local proxy that sets the mime-type correctly or maybe some javascript trickeries can help ? |
In #770 @wsnyder pointed out that it would be useful to have old reports published somewhere.
It appears the history is stored in the gh-pages at https://github.com/SymbiFlow/sv-tests/commits/gh-pages
We could do this by just always committing to a new directory and updating an "index" symlink, git will automatically dedup identical files in the repo. See the https://github.com/timvideos/HDMI2USB-firmware-prebuilt/tree/master/archive/master repository for example.
The text was updated successfully, but these errors were encountered: