Skip to content
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

Release 1.1.1 with only the GHC 7.10 fix #147

Closed
gelisam opened this issue Jun 30, 2015 · 18 comments
Closed

Release 1.1.1 with only the GHC 7.10 fix #147

gelisam opened this issue Jun 30, 2015 · 18 comments

Comments

@gelisam
Copy link
Owner

gelisam commented Jun 30, 2015

There is a lot of work scheduled for 1.2, let's not postpone this important fix until all that is implemented.

@gelisam
Copy link
Owner Author

gelisam commented Jun 30, 2015

The GHC 7.10 fix is #144. I don't want to merge that branch in though, since that would merge all the work on master which came before it. Cherry-pick? But then stable and master would diverge. Cherry-pick followed by merge into master?

@gelisam
Copy link
Owner Author

gelisam commented Jun 30, 2015

Looks like I need #123, too.

@gelisam
Copy link
Owner Author

gelisam commented Jun 30, 2015

And #140.

@gelisam
Copy link
Owner Author

gelisam commented Jun 30, 2015

With those modifications, compiles under GHC 7.10.1.

@gelisam
Copy link
Owner Author

gelisam commented Jul 1, 2015

Urh, but I can't figure out whether that version passes the tests, because of #141.

@gelisam
Copy link
Owner Author

gelisam commented Jul 1, 2015

Wait, Travis told me that the tests passed (with earlier versions of GHC). How is Travis running the tests?

@gelisam
Copy link
Owner Author

gelisam commented Jul 2, 2015

Like this:

> cabal-1.18 install --only-dependencies --enable-tests --enable-benchmarks
> cabal-1.18 configure --enable-tests --enable-benchmarks -v2
> cabal-1.18 build
> cabal-1.18 test

It's an improvement, since the tests now run... but they hang! They hang with ghc HEAD on Travis as well, so this is probably a ghc 7.10 regression.

@gelisam
Copy link
Owner Author

gelisam commented Jul 2, 2015

I vaguely remember having seen those tests hang before. In fact, isn't that why the doctests aren't being run in alphabetical order?

@gelisam
Copy link
Owner Author

gelisam commented Jul 5, 2015

It hangs in GHC.typecheckModule.

@gelisam
Copy link
Owner Author

gelisam commented Jul 5, 2015

The hang only occurs when I run the tests via cabal test, not when I run the test executable directly via ./dist/build/reference/reference.

@gelisam
Copy link
Owner Author

gelisam commented Jul 5, 2015

Oh, and the tests pass when I do it this way! I guess I could release this new version of Hawk, so that users of GHC 7.10 can install it from hackage, and then I'll investigate the hang some other time.

gelisam added a commit that referenced this issue Jul 5, 2015
For #147, I cherry-picked a few commits into stable, which caused
stable to diverge from master. This merge brings them back together.
@gelisam
Copy link
Owner Author

gelisam commented Jul 5, 2015

Merged into stable, tagged with 1.1.1, and merged back into master. Ready to release!

@gelisam
Copy link
Owner Author

gelisam commented Jul 6, 2015

Before releasing I wanted to run cabal-rangefinder to make sure that the version bounds for the dependencies were still good. Running it with ghc 7.10 caused cabal-rangefinder to reject many older versions of our dependencies, as those old versions are unsurprisingly not compatible with ghc 7.10. So I ran cabal-rangefinder again with ghc 7.6.3 and it confirmed that the version ranges were correct, except for transformers, which now requires >=0.4.1.0 instead of >=0.3.0.0. I wonder what change caused that?

@gelisam
Copy link
Owner Author

gelisam commented Jul 6, 2015

It actually compiles just fine with 0.3.0.0. Maybe cabal-rangefinder got confused because it failed on a version in between 0.3.0.0 and 0.4.1.0?

@gelisam
Copy link
Owner Author

gelisam commented Jul 6, 2015

Indeed, 0.4.0.0 is rejected by cabal because that specific version is being rejected by the indirect dependency transformers-compat.

@gelisam
Copy link
Owner Author

gelisam commented Jul 6, 2015

All this to say: the version ranges are good, I can release!

@gelisam
Copy link
Owner Author

gelisam commented Jul 6, 2015

I'm going through the process of uploading the package as a candidate release, and I'm discovering opportunities for improvements. I think the tagging and merging was premature, and the bidirectional merging especially will make the history hard to follow, so I have exceptionally decided to rewrite the history. So it you've fetched commits fc5808c and ed5f368 during the last 24 hours, sorry! Those commits no longer exist. Yes, I know that rewriting published history is frowned upon, but even though the repo is relatively popular (132 stars at the time of writing, yay!), my impression is that virtually no one except me is working on the code, so I doubt anybody fetched those commits.

gelisam added a commit that referenced this issue Jul 6, 2015
For #147, I cherry-picked a few commits into stable, which caused
stable to diverge from master. This merge brings them back together.
@gelisam
Copy link
Owner Author

gelisam commented Jul 6, 2015

Re-merged, re-tagged, released! Done.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant