freenode/#sbcl - IRC Chatlog
Search
20:53:33
pfdietz
There's a special case for the jump tables when all the results are constant forms, right?
21:12:02
pfdietz
That's the absolutely most recent git commit? Because stas walked back some of those initialization warnings.
21:15:31
Xach
sure. it's not just cl-store, though. i wish i tested earlier so that things could have been fixed before the release.
21:16:24
Xach
whenever closer adherence to the standard breaks stuff, it sometimes breaks stuff that is useful but not recently maintained
21:16:57
Xach
http://report.quicklisp.org/2019-11-26/failure-report/sel.html#software-evolution-library o noes
21:30:11
stassats
i could downgrade the defclass type mismatch to a style-warning, but i think there's little excuse to using mismatched initforms there
21:35:23
stassats
(FOMUS::DIV :TYPE (INTEGER 2) :ACCESSOR FOMUS::RULE-DIV :INITFORM 1 :INITARG :DIV)
22:39:39
Krystof
at some point between sbcl 0.8 and 0.9, we lost (I think by accident) the defstruct initform warning
22:44:19
Krystof
this is a thing that happens with Real Life: this was my last even vaguely-free evening to do a release before December. Sorry that it's causing a scramble at this point :-(
0:54:09
aeth
Krystof: On the other hand, I do this all of the time. I make an initform that's NIL when I want the user to have to provide it.
1:01:06
pfdietz22
If you want an initform for a slot (in a structure or standard class object) that forces the user to provide a value, use an error form. The initform is only evaluated (in either case) when the user does not provide a value.