Well done! Elements of the consensus (your section 2) seem to be * Cross-platform portability is very important, meaning: the same Haskell program should run on many platforms * The primary goal is to define a mid-level API, at the general level of Object I/O or GIO * The implementation technology (i.e. which underlying library is used) is not part of the specification. Indeed, it's a goal that the specification should constrain the implementation technology as little as possible. Specifically: (a) the specification should be such that it is possible to implement the spec with native look and feel (b) the specification should be such that a free-software implementation is possible I'd like to suggest a non-goal * The specification API should be Haskell 98 GUIs are an area where muti-parameter type classes and existentials are really useful, and it's a pity not to use them. Let's stick to features in the intersection of GHC and Hugs, though. Simon | -----Original Message----- | From: Axel Simon [mailto:A.Simon@ukc.ac.uk] | Sent: 11 February 2003 23:52 | To: gui@haskell.org | Subject: [GUI] after the opinion poll | | Hi all, | | I summarized everything that was said on this email list. I put the result | on the web and would like to ask everybody to review it and correct what I | got wrong: | | http://www.cs.ukc.ac.uk/people/staff/as49/poll.pdf | | After everyone has agreed to what they've said, I'd like to formulate some | sort of majority opinion and from that a mission statement. | | Thank you, | Axel. | | _______________________________________________ | GUI mailing list | GUI@haskell.org | http://www.haskell.org/mailman/listinfo/gui
On Wed, Feb 12, 2003 at 09:32:37AM -0000, Simon Peyton-Jones wrote:
Well done!
Elements of the consensus (your section 2) seem to be
* Cross-platform portability is very important, meaning: the same Haskell program should run on many platforms
* The primary goal is to define a mid-level API, at the general level of Object I/O or GIO
* The implementation technology (i.e. which underlying library is used) is not part of the specification. Indeed, it's a goal that the specification should constrain the implementation technology as little as possible. Specifically:
(a) the specification should be such that it is possible to implement the spec with native look and feel
(b) the specification should be such that a free-software implementation is possible
I'd like to suggest a non-goal
* The specification API should be Haskell 98
GUIs are an area where muti-parameter type classes and existentials are really useful, and it's a pity not to use them. Let's stick to features in the intersection of GHC and Hugs, though.
Thanks Simon! You safed my writing (and especially motivating) the summary. I will add this tonight. Axel.
On Wed, Feb 12, 2003 at 09:32:37AM -0000, Simon Peyton-Jones wrote:
Elements of the consensus (your section 2) seem to be
* Cross-platform portability is very important, meaning: the same Haskell program should run on many platforms ... * The implementation technology (i.e. which underlying library is used) is not part of the specification. Indeed, it's a goal that the specification should constrain the implementation technology as little as possible. Specifically:
(a) the specification should be such that it is possible to implement the spec with native look and feel ...
In the discussion, I saw considerable concern that these two goals might be incompatible, that it's not possible to provide really good native look and feel on multiple platforms. To address this, I saw this point being mentioned: * The specification should be extensible, so that non-portable Haskell programs can provide better native look and feel. Best, Dylan Thurston
participants (3)
-
Axel Simon -
Dylan Thurston -
Simon Peyton-Jones