freenode/#clim - IRC Chatlog
Search
7:33:03
beach
Sure. And I was just making conversation with him. He was standing by himself in a big room with all my colleagues present.
7:38:10
beach
All I do myself is an occasional sudoku, and I don't see how that teaches me any English.
7:38:20
ck_
Okay it depends on what we mean by 'learn' -- but from my experience the school-taught version of english for example is just horrible
7:42:33
ck_
Anyway. I'm thinking about whether or not I should plan to attend ELS next year. What do you say?
7:44:16
beach
100 or so participants. Many of which you know from IRC. And others that you have heard of.
7:44:21
ck_
Aah. I didn't question that -- it was more of a "should _i_ be there", "do I even qualify" type of question
7:44:50
beach
There is this widespread idea that conferences are about the papers, the talks, the research, etc.
7:45:19
beach
It's all about meeting people, talking to them during the breaks, getting to know them.
7:46:49
beach
You will meet jackdaniel, fe[nl]ix, Xof, heisig, ioa, Didier Verna, Jim Newton, perhaps even loke.
7:52:16
jackdaniel
ck_: me and my wife are going every year (4 times already) as a form of holiday to els (last year was skipped due to a new member of the family), we come a few days before and stay a few days after the symposium
7:58:37
ck_
If I can find the time to prepare something I'll ask you to look it over, then possibly submit it. Maybe over the christmas break. Would you do that?
8:03:26
jackdaniel
I'm at a point where I'm analyzing ECL's IR and I want to compare it with cleavir's
8:04:04
no-defun-allowed
(Small thing I noticed, should the example implementations of RPLACA/D have CONS instead of (OR CONS NULL) as the expected type?)
10:30:09
jackdaniel
beach: what you might find interesting ECL compiler c1forms resemble very closely concrete syntax tree as described in cleavir document (they serve more purposses additionaly though)
13:27:52
beach
jackdaniel: I see what you mean. Though c1forms seem to be created by the compiler as opposed to being consumed by it.
13:30:47
jackdaniel
beach: yes, in cleavir/eclector things are finely separted, in ecl many things are interwinded
13:31:57
jackdaniel
c1form also serves a purpose of ast and ir (and representation is pretty simplistic)
13:33:29
jackdaniel
I've mentioned that I'm making a document for better ecl compiler comprehension, here is its current state: https://gist.github.com/dkochmanski/fc361643488c8059a6030f143ac94a6d
13:34:47
beach
Great! Is it just for your personal consumption, or do you expect others to read it as well?
13:35:15
jackdaniel
for now it is a sketch for myself, I plan to write something better organized later
13:35:42
jackdaniel
if you have some remarks regarding content I'm happy to hear them, but I'm not interested in typos (I'd run flyspell first if I were)
13:36:44
jackdaniel
there are more to come, I didn't explain i.e environments, but I've already learned a lot while writing it
13:37:28
beach
Yes, I can see how attempting to write such a document would force you to understand things in detail.
13:38:18
beach
It is interesting to see how people write Common Lisp compilers when they don't have standard classes nor generic functions.
13:39:10
jackdaniel
dispatch tables are a clever substitute. I think that I'd find it much harder to follow the code if there were auxiliary methods for generic functions
13:40:07
jackdaniel
they are easy to write but I find them hard to read when trying to comprehand the full behavior
13:40:12
beach
Wouldn't the dispatch table just be primary methods with an EQL specializer the way we do in Cleavir?
13:40:59
jackdaniel
yes, they would. and ecl could use clos to its full extent, because it is compiled with full cl available
13:41:57
jackdaniel
bytecodes compiler and interpreter are C programs and they are used to first load Common Lisp, and then the compiler
13:43:41
jackdaniel
function eval is implemented in C (there is a single one). there are two compilers and a common runtime