-
Notifications
You must be signed in to change notification settings - Fork 4
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
removed lwt and updated pingtest #87
Conversation
Can you update travis-ci files to not build lwt? |
There are still some |
Pushed already. Let it rebuild. On Wed, Feb 5, 2014 at 10:49 AM, Spiros Eliopoulos <notifications@github.com
|
No. Those are used by the Async code too. On Wed, Feb 5, 2014 at 10:49 AM, Spiros Eliopoulos <notifications@github.com
|
I know you all like deleting code, but some folks are still using Lwt. why can't you hold-off on deleting this for another month or so? |
because it would be a pain to maintain. This is for the release in a month. On Wed, Feb 5, 2014 at 10:52 AM, Andrew Ferguson
|
you're already not maintaining it (it's deprecated) I'm asking you to simply not delete it yet, so we can continue building against master. |
ADF: sorry to be the bearer of bad news. But my opinion is if you're -N On Wed, Feb 5, 2014 at 10:57 AM, Andrew Ferguson
|
I agree it adds clutter. I'm confused why it adds complexity when it's already in a separate module guarded by an --enable-lwt flag. |
Better: let's discuss this on the call. My quick reaction is, doesn't -N On Wed, Feb 5, 2014 at 11:23 AM, Andrew Ferguson
|
sure, we can discuss on the call. forking something big like Lwt makes everyone happy only if the upstream maintainers don't want fixes from the fork, and the fork maintainers don't want fixes from upstream (unless they are prepared to do a lot more work with git). I would prefer not to be in that camp, but if you all do, then that's your call. it happened to us before with NetCore. |
Let's make sure we're all on the same page about why this is necessary, which involves repeating already-stated facts. As mentioned in #65, the lwt interface is deprecated. We no longer feel an obligation to maintain it, and we won't accept patches to it. Given it is no longer being maintained, the lwt sub-package should not make it into the next release and should therefore be deleted from repository before that time. The next release is planned for the end of the month. Until then, the lwt package can stay in the repository as long as it does not hinder development. How might it hinder development? If we change APIs that the lwt sub-package depends on, then it will likely no longer build. Even worse, it may build but exhibit incorrect behavior. To your point about git. Git is a cornucopia of esoteric features. It's why we love it. One slightly esoteric feature being merge strategies. But if that's not obscure enough, merge strategies take parameters. One such parameter is Once that's set up, you can merge and rebase from this repo as much as you please. For non-lwt patches that you'd like to contribute, that's as simple as creating a new branch off of this repo's master, cherry-picking some commits, and issuing a pull request. More or less, the usual workflow for open-source contributions. In sum:
|
thank you. that's all I needed to know. if you are already planning to change the abstraction layer that Lwt & Async both implement, then I totally understand. |
Ibid., p. same. |
sorry, I don't understand. what do you mean? (and yes, I know what those abbreviations mean on their own :-) |
We're on the same page now. |
thanks. I kept trying to figure out what previous citation you wanted to reference |
Not to beat a dead horse, but I just lost ~45 minutes each of the past two days dealing with async flags, failures in Travis, etc. This makes me sad, because I rarely get time to hack so I'd rather not be spending it fighting with configuration settings. Maybe I'm dumb or too out of date... but I also think that keeping lwt alive does add complexity and is slowing us (or at least me) down. |
sorry to hear that Nate. that sort of thing really sucks. :-( (just to be clear: I became totally fine with this going in once Spiros pointed out that y'all were planning to change the abstraction layer as part of the push to release NetKAT. LWT was going to break all on its own once that happened.) |
No worries. It's mostly my fault for letting my branch sit around for so many weeks. -N |
Oh to be living the #postdocballerlyfe instead. |
Conflicts: .travis.yml _tags lib/META setup.ml
removed lwt and updated pingtest
This removes openflow.lwt. There is no point maintaining it any more. I've also updated pingtest to use async instead of lwt. Pingtest now builds when the test flag is set (I removed the end_test flag). We need to always keep tests like these building.