4.19 KB
Newer Older
Ralf Jung's avatar
Ralf Jung committed
# Contributing to the Iris Coq Development
Ralf Jung's avatar
Ralf Jung committed

3 4 5
Here you can find some how-tos for various thing sthat might come up when doing
Iris development.

Ralf Jung's avatar
Ralf Jung committed
## How to submit a merge request

8 9 10 11 12
To contribute code, you need an [MPI-SWS GitLab account][account] (use the
"Register" tab).  Please send your MPI-SWS GitLab username to [Ralf Jung][jung]
to enable personal projects for your account.  Then you can fork the
[Iris git repository][iris], make your changes in your fork, and create a merge
Ralf Jung's avatar
Ralf Jung committed
13 14 15

Please do *not* use the master branch of your fork, that might confuse CI.  Use
a feature branch instead.

17 18 19 20

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
## How to update the std++ dependency

* Do the change in std++, push it.
* Wait for CI to publish a new std++ version on the opam archive, then run
  `opam update iris-dev`.
* In Iris, change the `opam` file to depend on the new version.
* Run `make build-dep` (in Iris) to install the new version of std++.
  You may have to do `make clean` as Coq will likely complain about .vo file

## How to write/update test cases

The files in `tests/` are test cases.  Each of the `.v` files comes with a
matching `.ref` file containing the expected output of `coqc`.  Adding `Show.`
in selected places in the proofs makes `coqc` print the current goal state.
This is used to make sure the proof mode prints goals and reduces terms the way
we expect it to.  You can run `make MAKE_REF=1` to re-generate all the `.ref` files;
38 39 40 41 42 43
this is useful after adding or removing `Show.` from a test.  If you do this,
make sure to check the diff for any unexpected changes in the output!

Some test cases have per-Coq-version `.ref` files (e.g., `atomic.8.8.ref` is a
Coq-8.8-specific `.ref` file).  If you change one of these, remember to update
*all* the `.ref` files.
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89

## How to measure the timing effect on a reverse dependency

So say you did a change in Iris, and want to know how it affects [lambda-rust]
or the [examples].  To do this, check out the respective project and change its
`.gitlab-ci.yml` to contain only one build job, which should look like
  <<: *template
    OPAM_PINS: "coq version 8.9.0 git git+"
    TIMING_CONF: "coq-8.9.0"
  - fp-timing
You will have to adjust this a bit: you should use the same Coq version as
whatever the master branch uses for its timing job, which you can determine by
checking its `.gitlab-ci.yml`.  If you change the Coq version, remember to do it
in both places (`OPAM_PINS` and `TIMING_CONF`).  You will also have to adjust
the Iris branch being used, which is determined after the `#` in `OPAM_PINS`.
If you are in doubt, ask on Mattermost *before* pushing your branch.

Once you are confident with your CI configuration, push this to a new branch
whose name starts with `ci/`.  It should usually be of the form
`ci/yourname/feature`.  You should see a pipeline running in GitLab with just a
single job, and you can follow its progress there.

When the job is done, you should be able to see it as a single dot on our
[statistics server][coq-speed] after selecting the right project and branch.
Click on "Coq-Speed" on the top-left corner to switch to another dashboard, and
select "Coq-Compare".  Now you can select the project and the two measurements
you want to compare, which would be the SHA of the commit you just created as
"Commit 2", and the SHA of its parent as "Commit 1".  Don't forget to also
select the right configuration for both of them.  The "Grouping" is a regular
expression that you can use to switch between per-file, per-directory and
per-project grouping of the measurements.

If you changed your Iris branch and want to make another measurement, *do not*
just "Retry" the CI job.  That will lead to an error, because you would end up
with two measurements for the same commit.  Instead, create an empty commit in
your branch of the to-be-measured project (`git commit --allow-empty -m
"rerun"`), and push that.