Hi,
During our first UK Smalltalk meeting, Marcel Weiher presented the ideas behind Higher Order Messages [1]. With them your collection code looks much nicer. Compare:
Without HOM:
#(1 2 3) collect: [:e | e + 1] #(1 2 3) inject: 0 into: [:a :b | a + b] #(1 2 3) addAll: #(4 5)
With HOM:
#(1 2 3) collect + 1 #(1 2 3) reduce + 0 #(1 2 3) add: #(4 5) all
Has anybody implemented them Squeak? Marcel mentioned a port to VW, but he also mentioned that it is quite easy to implement.
The APL guys at the meeting where very excited about the #reduce message (which they denote by / ). They didn't like that you have to specifiy the initial value. I think that with some double dispatching, we could avoid having that initial value. We could have code like this: #(1 2 3) / +
Cheers, Francisco
[1] http://www.metaobject.com/papers/Higher_Order_Messaging_OOPSLA_2005.pdf
Those are neat, but I find the originals more readable (especially for newcomers).
-C
p.s.
I'm enjoying stuff like "18 April 2006" a lot more. :) Quoth is full of things like that (e.g., "1 second, 3rd g sharp, 60 percent volume, 2nd channel").
Craig said:
"Those are neat, but I find the originals more readable (especially for newcomers)"
But you are not a newcomer!! How do you know they are more readable?
In Marcel's paper, there is some justification as to why the code with HOM is more readable. Without HOM, you have explicit iteration and when reading such expressions, you are forced to jump between two layers of abstractions. You have to think about the collection and about the elements of the collection. Too bad.
Taking the example of Marcel and Stef's paper, if Sally and Alice are two objects representing a manager with its employees, and Sally leaves the company, we need to add Sally's reports to Alice's. Your code would look like:
sally reports do: [ :each | alice addReport:each ]
The main subject of this expression (Alice) is now in the middle. You mentally jump inside and outside of the block. Of course, experienced Smalltalkers do this without noticing. But it would be more natural to express the same thing with HOM:
alice addReport: sally reports all
Beautiful, isn't it?
-Francisco
----- Original Message ----- From: "Craig Latta" craig@netjam.org To: squeak-dev@lists.squeakfoundation.org Sent: Tuesday, April 18, 2006 11:47 PM Subject: Re: HOM - Higher Order Messages
-C
p.s.
I'm enjoying stuff like "18 April 2006" a lot more. :) Quoth is full of things like that (e.g., "1 second, 3rd g sharp, 60 percent volume, 2nd channel").
-- Craig Latta improvisational musical informaticist www.netjam.org Smalltalkers do: [:it | All with: Class, (And love: it)]
Francisco Garau wrote:
alice addReport: sally reports all
Beautiful, isn't it?
Unusual for sure. Beautiful? I don't know. The statement "sally reports all" means something entirely different in english than in it does in the above (which is largely because the above violates basic english language structure making "reports" a noun where in english it can only be a verb). If we take it from plain english we might get something like "alice adds each of sally's reports" which we might translate as "alice add: each_of sally reports" or so but that's not exactly beautiful either. Unusual, for sure ;-)
Cheers, - Andreas
Those are neat, but I find the originals more readable (especially for newcomers)"
But you are not a newcomer! How do you know they are more readable?
I teach newcomers.
-C
On 19-Apr-06, at 8:33 AM, Craig Latta wrote:
Those are neat, but I find the originals more readable (especially for newcomers)"
But you are not a newcomer! How do you know they are more readable?
I teach newcomers.
Don't you mean Newcomers allStudents do:[:newcomer| Craig teach: newcomer]
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: BOZO: Use Multics operating system
On 20-Apr-06, at 5:21 AM, Lothar Schenk wrote:
tim Rowledge wrote:
On 19-Apr-06, at 8:33 AM, Craig Latta wrote:
I teach newcomers.
Don't you mean Newcomers allStudents do:[:newcomer| Craig teach: newcomer]
Why should the newcomers have to do anything, if Craig does the teaching?
Well in *my* day (me being so old and creaky that my tutors were pre- hominid) the students were expected to do a great deal of work. To the point of burning out about 40% by the middle of second year and another 10% or so before the commencement of third year.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Useful Latin Phrases:- Estne volumen in toga, an solum tibi libet me videre? = Is that a scroll in your toga, or are you just happy to see me?
Craig: "I'm enjoying stuff like "18 April 2006" a lot more. :)"
Grin. By they way, "18 April 2006" just happens to be the 100-year anniversary of the Great San Francisco Earthquake. Hope that doesn't "shake you up." :-)
So far, my main focus with Chronos has been correctly modeling and implementing dates and times, and providing a rich time model.
Adding "syntactic sugar" can be done later. And of course, the "method-selector-competitition among libraries" issue must be considered, whenever convenience methods are added, since they usually involve adding methods to fundamental classes of the base Smalltalk class library.
--Alan
Hello, I have a package on SqueakMap called "LazyCollections". The idea originally was to implement the common collection protcols to lazy evaluate. But, when I read the HOM paper, I decided to try it out. It turned out to be trivial to implement in "LazyCollections".
For an example, see LazyCollectionTest>>testIterator. The LazyIterator class is where the magic begins. I didn't implement everything in the HOM paper. I mainly did LazyCollections as an experiment. If you get anything out of it, please let me know. ------------------------------------------------ Blaine Buxton, Mad Scientist In Training http://www.blainebuxton.com
From: "Francisco Garau" francisco.garau@gmail.com Reply-To: The general-purpose Squeak developers listsqueak-dev@lists.squeakfoundation.org To: "Squeak List" squeak-dev@lists.squeakfoundation.org Subject: HOM - Higher Order Messages Date: Tue, 18 Apr 2006 22:25:40 +0100
Hi,
During our first UK Smalltalk meeting, Marcel Weiher presented the ideas behind Higher Order Messages [1]. With them your collection code looks much nicer. Compare:
Without HOM:
#(1 2 3) collect: [:e | e + 1] #(1 2 3) inject: 0 into: [:a :b | a + b] #(1 2 3) addAll: #(4 5)
With HOM:
#(1 2 3) collect + 1 #(1 2 3) reduce + 0 #(1 2 3) add: #(4 5) all
Has anybody implemented them Squeak? Marcel mentioned a port to VW, but he also mentioned that it is quite easy to implement.
The APL guys at the meeting where very excited about the #reduce message (which they denote by / ). They didn't like that you have to specifiy the initial value. I think that with some double dispatching, we could avoid having that initial value. We could have code like this: #(1 2 3) / +
Cheers, Francisco
[1] http://www.metaobject.com/papers/Higher_Order_Messaging_OOPSLA_2005.pdf
_________________________________________________________________ Dont just search. Find. Check out the new MSN Search! http://search.msn.click-url.com/go/onm00200636ave/direct/01/
squeak-dev@lists.squeakfoundation.org
-
Alan Lovejoy -
Andreas Raab -
Blaine Buxton -
Craig Latta -
Darius Clarke -
Francisco Garau -
Lothar Schenk -
tim Rowledge