freenode/#sicl - IRC Chatlog
Search
9:24:00
heisig
ACTION just returned from teaching Common Lisp to a group of very motivated students.
9:40:28
heisig
beach: Good question. I think it is just because they notice I am very enthusiastic about it. And I promised them it will make their life easier and their programs better.
9:50:40
fiddlerwoaroof
https://pdfs.semanticscholar.org/7e49/0023f84845c750f4a04b7dca8a0225d3e5d1.pdf
10:07:20
Shinmera
complexity analysis and proofs of sorts is one of the first things we did in our algorithms lecture.
10:13:58
no-defun-allowed
heisig: somewhat irrelevant, I got my music exam back and I got a C. I won't complain, but I got higher marks on my other assessments but the C goes on my report.
10:14:47
no-defun-allowed
Bucket sort also scales with how many digits your representation requires and I doubt the space complexity is low. It's quite amusing nonetheless.
10:16:16
no-defun-allowed
fiddlerwoaroof: they are on the fritz. Bogosort is the best sort nonetheless.
10:22:19
no-defun-allowed
Bucketsort probably wouldn't write a good CL:SORT as far as I'm aware, as the function takes a comparing Boolean functions.
10:36:21
fiddlerwoaroof
Yeah, I think people in the Haskell community have figured out how to generalize it
12:48:33
beach
HAH, SLOT-VALUE-USING-CLASS is a generic function that does not fit the general bootstrapping scheme I have implemented. In this scheme, there is an assumption that the arguments of a generic function are all of the same let's call it "category". Either they are all host object, all bridge objects, or all ersatz objects. But SLOT-VALUE-USING-CLASS takes an object as its second argument, and the class of that object as its first
12:48:33
beach
argument. So those two arguments are necessarily in different categories during bootstrapping.
12:49:20
beach
It's not a big deal, of course. I can use the non-generic versions of SLOT-VALUE-USING-CLASS, and (SETF SLOT-VALUE-USING-CLASS) during bootstrapping.
15:58:24
beach
The only major difficulties in phase 4 were due to my messing up the bootstrapping code in phase 3. As I predicted, phases 3 and 4 were very similar.
15:59:14
beach
I am going to call it a day for the purpose of bootstrapping. I'll just hang out on IRC, fix dinner, etc. Tomorrow, I'll try to "tie the knot".
16:08:11
jcowan
beach: possible subtypes of sequence to consider: queue, deque, doubly linked list, compactly stored numeric range
16:11:13
beach
I haven't thought much about whether I should make the sequence functions generic, or do what SBCL does, i.e. trampoline to a generic function when neither a list nor a vector is given.
16:35:40
splittist
A simple matter of programming! If this was a pure research project you would now publish your papers and abandon the code. Fortunately for us, you have loftier goals.