A babel plugin adding the ability to rewire modul dependency. This enables to mock modules for testing purposes.
People
Repo info
Activity
speedskater
@speedskater
In the mean time we could figure out how to approach es6 or even es7 apis as well as support for different code coverage utils (by providing some kind of plugin api)
I think this will take some more discussions as it is not straight forward and each approach has its unique pros and cons.
But i think it will definitively make sense to create new issue for discussing this approaches.
Eli White
@TheSavior
Agreed
speedskater
@speedskater
What do you think about this es6, es7 problems?
Eli White
@TheSavior
It would help for us to outline all the different problems that a solution needs to solve
I think that is the only way to sufficiently validate different solutions
speedskater
@speedskater
sounds good. Therefore we need to create samples for all this problems, which are not tackled till now in our tests.
Eli White
@TheSavior
I don’t think it is worthwhile to tackle that before getting out a stable babel 6 release
speedskater
@speedskater
So I will propose the following: till the end of this week, I will release a new version and close babel6 support. Furthermore I will create a new issue which should be used for collecting ideas and approaches for future work.
Eli White
@TheSavior
Since we won’t specifically know what features will be supported and which won’t in the first version and then what we need to tackle as a follow up
Sounds good, thanks
speedskater
@speedskater
On monday the 14th March. I will create a new branch and start with the first samples which currently do not work which come to my mind. We will than collect new ideas till mid may and start after that with the implementation of this approach. In the mean I will try to fix smaller issues and release bug fix release on every third monday starting with the 14th of March.
Do you have any further suggestions regarding this roadmap?
Eli White
@TheSavior
I think any expectation of a plan that far out with the few details we have seems likely to be inaccurate to some level.
speedskater
@speedskater
thats definitively true, but we can at least reduce the expectations.
Eli White
@TheSavior
I think the part that is reasonable is your belief on how often you are able to work on it.
speedskater
@speedskater
that's true
Eli White
@TheSavior
If people have problems with that (myself included), then they should dig into the code and solve things themselves.
speedskater
@speedskater
;)
Eli White
@TheSavior
But we should have a way to support people who are able to do so
speedskater
@speedskater
definitely.
It is important and we should have some kind of workflow when merges and releases will occur.
Eli White
@TheSavior
I imagine it is very demoralizing to put out a PR that passes builds and looks good but is told it won’t be merged until your every three week period
_
speedskater
@speedskater
thats true.
Eli White
@TheSavior
I think merges and releases should be as quick as possible assuming that we have high confidence in the change
speedskater
@speedskater
yes thats true. It should be doable as soon as this babel 6 huge issue is out.
but I think we should have some kind of stages. I we merge fast we should provide a beta first.
Eli White
@TheSavior
What does that optimize for?
speedskater
@speedskater
for higher stability
Eli White
@TheSavior
Do we not have confidence in the test suite?
speedskater
@speedskater
we do, but I am not 100% that we have already tackled all use cases
Eli White
@TheSavior
What a beta implies is that we don’t trust our test suite and need some amount of manual testing
One approach is to improve our test suite with some amount of integration tests, possibly by doing something crazy like cloning jquery (or some large es6 package) and trying to rewire something in that
I don’t think that is a good long term solution though
I think it is reasonable to say that we support everything our unit tests cover
By merging and releasing quickly we enable it to be quicker to get fixes, and we at least validate against regressions
speedskater
@speedskater
thats reasonable but I mean more something like the wildcard imports we haven't considered in the first place but are out there. But I like your approach regarding using existing libraries and their unit tests for integration
we can do this. I would propose the following. I create the babel 6 release till the end of this week. than I will make you contributor as well and we try both to merge asap and release after some kind of four eye check?
Eli White
@TheSavior
I don’t have the context for specific code changes as you do
But that sort of plan sounds good in the aggregate
speedskater
@speedskater
Okay. Than I'will try my best to stick to the plan. I will have to go offline now. We can chat again in some days after the released version.
Eli White
@TheSavior
:+1:
speedskater
@speedskater
bye
speedskater
@speedskater
its done babel 6 version shipped as rc1. As soon as no major issues occur we will finally ship 1.0.0
and from there move on as planned
Andrew Throener
@trainerbill
@speedskater I am interested in testing out this plugin. I npm to the babel6-support branch and added it to babelrc and getting:
SyntaxError in plugin 'gulp-babel' Message: Unexpected token import
Andrew Throener
@trainerbill
changed to master branch and getting:
TypeError in plugin 'gulp-babel' Message: Filename must be a string
speedskater
@speedskater
@trainerbill I think you can just use the main branch and run 1.0.0-rc-1 as this is the current default version. the babel6-support branch is quite old
Eli White
@TheSavior
@speedskater I’m pretty excited about this version but it looks like it doesn’t quite work with our system at work yet