I was inspired by my recent dust-up with Pascal C to take a new tack on
Cells proselytization, ironically in line with what I already meant to
be the theme of my talk until someone said I should talk about Cells
(twist my arm!). It turns out I should have done both.
The idea behind Cells (spreadsheet for CLOS instances) is so simple no
one understands what one would do with it unless they already are. So I
should have just talked about how they serve some larger purpose,
application or library. I took a swipe at that in what is referred to as
"PDF #1" in my listing here: http://www.weitz.de/eclm2008/
PDF #2 might be fun for folks who already understand the power of this
type of library as a class and are curious about functional specifics of
this instance.
The two Google videos are a short opening rant and then a longer piece
about RDF, Cello, the Lisp Way, four girls in bikinis, and a surprising
amount on driving.
Meanwhile, I have been in a funk ever since Pascal failed to take up my
challenge and sup****t his claim that Cells was an untrustworthy train
wreck of code -- I was looking forward to some serious fun, especially
because the code /is/ terrifying. But I /think/ that is because I just
leave in place all the debugging stuff (disabled, of course) and because
there are a lot of sanity checks in there. I have taken those out before
only to put them back in when the next hard bug came up so I just leave
them in there, it is not as if I ever look at that code. Anyway, it
looks like I will have to play the part of PC if I want to have a brawl,
so... this code is horrible!:
(defun calculate-and-set (c)
(flet ((body ()
(when (c-stopped)
(princ #\.)
(return-from calculate-and-set))
#-its-alive!
(bwhen (x (find c *call-stack*)) ;; circularity
(unless nil ;; *stop*
(let ()
(inspect c)
(trc "calculating cell:" c (cr-code c))
(trc "appears-in-call-stack (newest first): "
(length *call-stack*))
(loop for caller in (copy-list *call-stack*)
for n below (length *call-stack*)
do (trc "caller> " caller #+shhh (cr-code caller))
when (eq caller c) do (loop-finish))))
(setf *stop* t)
(c-break ;; break is problem when testing cells on some CLs
"cell ~a midst askers (see above)" c)
(error "see listener for cell rule cycle diagnotics"))
(multiple-value-bind (raw-value propagation-code)
(calculate-and-link c)
(when (and *c-debug* (typep raw-value 'cell))
(c-break "new value for cell ~s is itself a cell: ~s.
probably nested (c? ... (c? ))"
c raw-value))
(unless (c-optimized-away-p c)
; this check for optimized-away-p arose because a rule
using without-c-dependency
; can be re-entered unnoticed since that clears
*call-stack*. If re-entered, a subsequent
; re-exit will be of an optimized away cell, which we
need not sv-assume on... a better
; fix might be a less cutesy way of doing
without-c-dependency, and I think anyway
; it would be good to lose the re-entrance.
(md-slot-value-assume c raw-value propagation-code)))))
(if (trcp c) ;; *dbg*
(wtrc (0 100 "calcnset" c) (body))
(body))))
Nonsense! Here is the same thing without comments or debugging/defensive
stuff!:
(defun calculate-and-set (c)
(multiple-value-bind (raw-value propagation-code)
(calculate-and-link c)
(unless (c-optimized-away-p c)
(md-slot-value-assume c raw-value propagation-code))))
Wow. Too bad I did not abbreviate:
(defun calc-n-set (c)
(multiple-value-bind (v prop)
(calc-n-link c)
(unless (c-optmzd? c)
(md-slot-v-assume c v prop))))
Damn.
Of course the run-time diagnostic on cycles is a courtesy to Cells users
(including moiself) so it should stay, but I could throw an error
condition and let /that/ offer the debug diagnostics, keeping the code
itself clean for when the white glove crowd comes snooping.
But now it is time for Algebra.
kenny
--
http://smuglispweeny.blogspot.com/
http://www.theoryyalgebra.com/
"I've never read the rulebook. My job is to catch the ball."
-- Catcher Josh Bard after making a great catch on a foul ball
and then sliding into the dugout, which by the rules allowed the
runners to advance one base costing his pitcher a possible shutout
because there was a runner on third base.
"My sig is longer than most of my articles."
-- Kenny Tilton
|
