freenode/#sicl - IRC Chatlog
Search
6:54:03
beach
It sounds like you can access memory that is located on a different computer as if it were local. But I haven't read the details.
6:57:35
beach
You need to be patient. Right now, I am working on this year's ELS submissions. Then I'll be back on code generation.
7:02:10
ebrasca
I think if you see some of my code , you are going to understand I am not very patient.
7:18:08
beach
It is in the first sentence in the first chapter of the specification. "sickle", as in the tool for cutting grass.
16:07:31
beach
jackdaniel: I did the following experiment in ECL: (defgeneric bla (x) (:method-combination and :lskdjflskjfd)) and then (defmethod bla and (x) (> x 10)) and finally (bla 20).
16:07:57
beach
No error was signaled, because ECL only checks whether the option in :MOST-SPECIFIC-LAST.
16:08:46
beach
This is because the short form is rewritten to the long form and the generated long form does not check that the option is valid. It only compares it to :MOST-SPECIFIC-LAST.
16:10:13
beach
Bike: creating a new method combination instance for each generic function the way ECL and Clasp does it means that if the DEFINE-METHOD-COMBINATION form is reevaluated with some modifications, the modification will not affect existing generic functions.
16:18:33
beach
While working on this paper, I keep thinking that it would be preferable if code like this could be shared between implementations. And often the reason it can't be is that each implementation is bootstrapped differently.
16:18:34
beach
But then I was thinking whether it really matters how (say) ECL is bootstrapped. If not, it could be bootstrapped like SICL is, except that C code would be generated rather than an executable.