On Thu, Nov 25, 2010 at 10:26 PM, Mitar wrote:

Hi!

On Thu, Nov 25, 2010 at 8:39 PM, Antoine Latter wrote:

...

I was confused by your other thread - should Chans test as equal if one is duplicated from the other? With your current patch they do not (I haven't tested it, though, only looked at it).

I think it is OK that they are not equal if duplicated. Both MVar equality and Chan equality is more or less just reference equality.

Thanks for the proposal Mitar! To avoid confusion I've added an updated patch which adds the following note to dupChan: "(Note that a duplicated channel is not equal to its original. So: @fmap (c /=) $ dupChan c@ returns @True@ for all @c@.)" Regards, Bas P.S. What about a short deadline of a week for this proposal?

On Thu, Nov 25, 2010 at 10:40 PM, Mitar wrote:

On Thu, Nov 25, 2010 at 10:35 PM, Bas van Dijk wrote:

...

What about a short deadline of a week for this proposal?

I am/was not really sure what you meant by that?

Library proposals should have a timescale for consideration (to focus the community's attention for a limited period). IMHO a deadline of a week is enough for this proposal because I think it's trivial. At the end of the discussion period, if consensus was achieved, set the ticket state to "patch" so that someone will commit it for you. Bas

On Thu, Nov 25, 2010 at 10:59 PM, Mitar wrote:

On Thu, Nov 25, 2010 at 10:50 PM, Bas van Dijk wrote:

...

Library proposals should have a timescale for consideration (to focus the community's attention for a limited period). IMHO a deadline of a week is enough for this proposal because I think it's trivial.

Where do I announce/set this timescale? Just tell/propose that it is a week?

Yes that should be enough. It would be nice to also include it in the patch though. Bas

Hi!

Yes that should be enough. It would be nice to also include it in the patch though.

OK. So: I am proposing a week to discus things. Starting now. Mitar

On Fri, Nov 26, 2010 at 11:40 PM, Mitar wrote:

And SampleVar could also be an instance of Eq

It already is.

and Typable.

Good point! Proposal with patch: http://hackage.haskell.org/trac/ghc/ticket/4537 Let's also discuss (not that there's much to discuss) this proposal in this thread. Regards, Bas

On Fri, 2010-11-26 at 00:03 +0100, Mitar wrote:

Hi!

...

Yes that should be enough. It would be nice to also include it in the patch though.

OK. So:

I am proposing a week to discus things. Starting now.

Seems a perfectly reasonable proposal. Given that IO variables such as IORef, MVar, TVar etc are all in Eq, then other mutable objects with reference semantics such as Chan can also reasonably be in Eq. What about QSem and QSemN? The same reasoning apply to them doesn't it? Duncan