tag:blogger.com,1999:blog-11295132.post215149046042848636..comments2008-03-15T15:36:47.782-07:00sigfpehttp://www.blogger.com/profile/08096190433222340957noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-11295132.post-49440542942509012972008-03-15T15:12:00.000-07:002008-03-15T15:12:00.000-07:00re: your thinking about composing comonads -- excellent, as I've observed that you thinking about classes of problems results in solutions on your blog.<BR/><BR/>As to attribute grammars for Haskell, have you looked at them? Applied them? Do you see them being useful in these problem domains, or do they address an entirely different class of problems?geophfhttp://www.blogger.com/profile/09936874508556500234noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-48807242049225864572008-03-15T07:16:00.000-07:002008-03-15T07:16:00.000-07:00geophf,<BR/><BR/>By coincidence, I've been thinking about stacking comonads for the last couple of days. Not got very far yet. A web search reveals almost no hits on "comonad transformer".sigfpehttp://www.blogger.com/profile/08096190433222340957noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-36827386070998964352008-03-15T06:00:00.000-07:002008-03-15T06:00:00.000-07:00I've been reading your explorations of comonads -- your explanations using examples of cellular automata and image processing have been very helpful for me to understand them a bit better. In parallel, I've been reading about Haskell <A HREF="http://www.haskell.org/tmrwiki/WhyAttributeGrammarsMatter" REL="nofollow">attribute grammars</A>. Do comonads address a class of problems well that attribute grammars do not? <BR/><BR/>One of the issues with monads, even with transformers, is that they "stack" poorly (the result depends on order of composition and each monad's internal implementation detail); do comonads compose better than monads?geophfhttp://www.blogger.com/profile/09936874508556500234noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-44339024878691070702007-02-09T13:42:00.000-08:002007-02-09T13:42:00.000-08:00(I guess kotsu=Nick Frisby.)<BR/><BR/>Of course! cfix and loeb deal with relative and absolute references. I hadn't thought of it that way. It's truly bizarre how fiddling about semi-randomly with types leads to new ways to view concepts that are already well known and considered natural.<BR/><BR/>I'll have a look at Uustalu and Vene's LVS. Their recent papers have been really interesting.sigfpehttp://www.blogger.com/profile/08096190433222340957noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-55131715406302267092007-02-09T13:32:00.000-08:002007-02-09T13:32:00.000-08:00Forgot to sign! Nick Frisby here again :)kotsuhttp://www.blogger.com/profile/18174505241754470567noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-22849866810589248972007-02-09T13:31:00.000-08:002007-02-09T13:31:00.000-08:00That's a nice contrast you've developed.<BR/><BR/>My first post was inspired mostly by type signatures. Now that I've looked at your actual definition of loeb :), I readily see the difference you're highlighting. As you pointed out, loeb does not need the comonadic structure. This highlights the big difference: comonadic structure (and thus cfix) caters to relative references [1] whereas loeb allows for absolute references.<BR/><BR/>In your spreadsheet example, you use !! to get the nth cell in the list. A comonadic approach would allow a cell to refer to "the cell to my left" without needing to know its own cell number. Both features are supported in various ways by spreadsheets: whenever I do a fill-down in Excel, every now and then I'm surprised that the relative references worked how I had intended.<BR/><BR/>Just a couple of notes: to get cfix to "return a whole list" you can use cobind cfix. Also, to have access to the "entire output list", you can use a more robust comonad (like Uustalu and Vene's LVS) with an absolute access function a la !!.<BR/><BR/>Thanks for the discussion, that was a nice contrast to appreciate.<BR/><BR/>[1] - This is why zippers induce comonads; I can talk about neighbors in various directions relative to wherever I'm currently at in the structure.kotsuhttp://www.blogger.com/profile/18174505241754470567noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-90741346051068419282007-02-05T09:53:00.000-08:002007-02-05T09:53:00.000-08:00kea,<br /><br />If you think some of the jargon looks like gobbledygook, you should check out the published papers that some of this stuff comes from :-) It's been hard work trying to make inroads into this stuff over the last two years.sigfpehttp://www.blogger.com/profile/08096190433222340957noreply@blogger.comtag:blogger.com,1999:blog-11295132.post-46718312853971743342007-02-03T15:55:00.000-08:002007-02-03T15:55:00.000-08:00Your blog continues to amaze me, and I just wanted to say <b>thanks</b>. Most of the CompSci jargon is gobbledygook to me, but I really appreciate the effort you're taking to talk about monads and the like.Keahttp://www.blogger.com/profile/05652514294703722285noreply@blogger.com