Hotfix

[jdegoes]

An FFI function had an error that wasn't detected by the build. Fixed the error, also added jsvalidate as an NPM dependency and incorporated it into the build process so the build won't complete if there are any FFI errors.

I'd suggest pushing this hotfix to the 0.3.0 tag directly, if possible. Review by @garyb etc.

[paf31]

How did this not get caught by core-tests?

[paf31]

Merged, and re-released as 0.3.0. Please be sure to run core-tests for any core library changes in future.

[paf31]

Oh I see what happened. Dead code elimination removed this function because it wasn't being tested.

[jdegoes]

Yep. Didn't detect it until I was updating purescript-quickcheck. :-/

PR coming shortly for core-tests to add some tests and update dependencies / fix versions.

[paf31]

Cool, thank you.

[jdegoes]

Maybe it's bower being weird, but I'm still getting the "old" 0.3.0. Any ideas?

[paf31]

bower cache clean?

[paf31]

It's not working for me either, but the tag clearly includes the fix.

[jdegoes]

bower cache clean did the trick for me. Well, I also had to delete output and bower_components. It appears there are several layers of caching.

[joneshf]

Also ran into this, should probably also push a 0.3.1 tag, since anything that depends on this (lots of libs) would've pulled the bad 0.3.0 tag and will now need a cache clean until some greater tag is released. It seems like if someone installed/searched/updated almost any purescript library (transitive deps) they'll pull the bad tag since bower will cache this when it looks at the dep. Like, if you did bower install purescript-dom before the tag was updated, it pulled the bad tag into cache when it looked at this repo. At least that's the best I can imagine why I got bad cache w/o explicitly dealing with this lib.

[jdegoes]

Agreed, can't hurt to push a new patch fix just in case.