<version> is a standard header with C++20 which could cause issues if a library checks it exists then imports it
Because we have the root of this repo in the include path, it'd see that <version> exists (with, e.g., __has_include), and then try to include it as a header
But because it's just a file that says 0.1, this would fail
This happens with libpqxx 7
Without this commit, two jobsets using the same repository as input,
but different `deepClone` options, end up incorrectly sharing the same
"checkout" for a given (`uri`, `branch`, `revision`) tuple. The
presence or absence of `.git` is determined by the jobset execution
order.
This patch adds the missing `isDeepClone` boolean to the cache key.
The database upgrade script empties the `CachedGitInputs` table, as we
don't know if existing checkouts are deep clones. Unfortunately, this
generally forces rebuilds even for correct `deepClone` checkouts, as
the binary contents of `.git` are not deterministic.
Fixes#510
Nixpkgs on unstable has removed `stdenv.lib` as they've been warning for a while now. This removes the extra references to it in the flake.nix
I'm not entirely sure if this is right, but I figured it was trivial enough to give a quick try using the GH Editor while I was waiting for a job to finish
I broke this when I added `me.` in f1e75c8bff
I added me. to disambiguate `id`, but:
* eval.id works on the per-build page
* me.id works on the other pages
* Just id works everywhere if I drop:
, prefetch => { evaluationerror => [ ] },
but this causes a query per row to collect the evaluationerror
records later, this becomes significantly slow on non-trivial
datasets.
Using evals->current_source_alias will use the correct alias
whether it is me or eval or something else.
Exposes metrics:
* http_request_duration_seconds_bucket
* http_request_size_bytes_bucket
* http_response_size_bytes_bucket
* http_requests_total
with labels of action and controller to help identify popular
endpoints and their performance characteristics.
