Rationale
---------
Currently, tests are hard to discover. For instance, someone updating
`dovecot` might not notice that the interaction of `dovecot` with
`opensmtpd` is handled in the `opensmtpd.nix` test.
And even for someone updating `opensmtpd`, it requires manual work to go
check in `nixos/tests` whether there is actually a test, especially
given not so many packages in `nixpkgs` have tests and this is thus most
of the time useless.
Finally, for the reviewer, it is much easier to check that the “Tested
via one or more NixOS test(s)” has been checked if the file modified
already includes the list of relevant tests.
Implementation
--------------
Currently, this commit only adds the metadata in the package. Each
element of the `meta.tests` attribute is a derivation that, when it
builds successfully, means the test has passed (ie. following the same
convention as NixOS tests).
Future Work
-----------
In the future, the tools could be made aware of this `meta.tests`
attribute, and for instance a `--with-tests` could be added to
`nix-build` so that it also builds all the tests. Or a `--without-tests`
to build without all the tests. @Profpatsch described in his NixCon talk
such systems.
Another thing that would help in the future would be the possibility to
reasonably easily have cross-derivation nix tests without the whole
NixOS VM stack. @7c6f434c already proposed such a system.
This RFC currently handles none of these concerns. Only the addition of
`meta.tests` as metadata to be used by maintainers to remember to run
relevant tests.
It would be reasonable to have a Ruby program that depends on some other
program being in the PATH. In this case, the obvious thing to do would
be something like this:
bundlerApp {
# ...
buildInputs = [ makeWrapper ];
postBuild = ''
wrapProgram "$out/bin/foo" \
--prefix PATH : ${lib.makeBinPath [ dep ]}
'';
}
However, this doesn't work, because even though it just forwards most of
its arguments to `runCommand`, `bundlerApp` won't take a `buildInputs`
parameter. It doesn't even specify its own `buildInputs`, which means
that the `scripts` parameter to `bundlerApp` (which depends on
`makeWrapper`) is completely broken, and, as far as I can tell, has been
since its inception. I've added a `makeWrapper` build input if the
scripts parameter is present to fix this.
I've added a `buildInputs` option to `bundlerApp`. It's also passed
through to bundled-common because `postBuild` scripts are run there as
well. This actually means that in this example we'd end up going through
two layers of wrappers (one from `bundlerApp` and one from
bundled-common), but that has always been the case and isn't likely to
break anything. That oddity does suggest that it might be prudent to
not forward `postBuild` to bundled-common (or to at least use a
different option) though...
FWIW, as far as I can tell no package in nixpkgs uses either the
`scripts` or `postBuild` options to `bundlerApp`.
repstopdf is supposed to be a symlink to epstopdf. Then epstopdf looks
at "$0" to detect when restricted mode needs to be enabled. Unfortunately
our wrapper will drop all intermediate symlinks, which messes up "$0".
Restricted mode appears to be a security feature, so a test is
introduced to verify that the wrapper works as expected.
