This is a new step in my crusade against the braindead fad of starting
PR titles with a word that is completely redundant with github labels,
thus wasting prime first-line real-estate for something that isn't
necessary.
I noticed that every single one of these PRs are low-quality AI-slop, so
I think there is a strong case to be made for these PRs to be
auto-closed. A message is added before closing the PR, redirecting to
our contribution guidelines, so I expect quality first-time contributors
to read them and reopen the PR. In the case of spam PRs, the author is
unlikely to revisit a given PR, and so auto-closing might have a
positive impact. That's an experiment worth trying, imo.
A new pointless fad appeared recently where people just create a fairly
low information tag at the beginning of their github PR titles.
Something like `feat` or other keywords.
This seems to originate from the angular community and to be used for
automation scripts over there. We do not use any of those scripts and if
we did we would be using the github labels, which offer strictly
equivalent functionalities without wasting useful PR title space.
In order for these keywords to fail the validation, I am adding a check
that these directories listed indeed exist in the repository.
This adds two new CI targets. One is for building all supported keeper
executables, the other is for running unit tests on 32-bit Linux.
---------
Co-authored-by: Felix Lange <fjl@twurst.com>
This PR does a few things:
- Sets the gh actions runner sizes for lint (s) and test (l) workflows
- Runs the tests on gh actions in parallel
- Skips fetching the spec tests when unnecessary (on windows in
appveyor)
- Removes ubuntu appveyor runner since it's essentially duplicate of the
gh action workflow now
The gh test seems to go down from ~35min to ~13min.
Full disclosure: this has been generated by AI. The goal is to have a
quick check that the PR format is correct, before we merge it. This is
to avoid the periodical case when someone forgets to add a milestone or
check the title matches our preferred format.
This change enables more tests to run on GitHub actions. First, it
removes the `-short` flag passed to `go test`, unskipping some longer
running tests. We also enable the full consensus tests to run by
enabling submodules during git clone.
The EF now operates org wide runners with the `self-hosted-ghr` label.
These are auto-scaling runners which should ideally allow us to process
any amount of testing load we throw at them. The new runners have `HOME`
configured differently from the actual user home directory, so our
internal test for resolving `~` had to be adapted to work in this scenario.
This changes the go mod tidy check to use the go mod tidy -diff command,
removing the custom diffing for go.mod. The check for go.mod/go.sum is now
performed in the check_generate action.
Also included is a change where check_generate and check_baddeps will now
run on the GitHub Actions lint step.
---------
Co-authored-by: Felix Lange <fjl@twurst.com>
This PR updates the version of go used in builds and docker to
1.23.0. Release notes: https://go.dev/doc/go1.23
More importantly, following our policy of maintaining the last two
versions (which now becomes 1.23 and 1.22), we can now make use of
the things that were introduced in 1.22: https://go.dev/doc/go1.22
Go 1.22 makes two changes to “for” loops.
- each iteration creates new variables,
- for loops may range over integers
Other than that, some interesting library changes and other stuff.
