libera/#sicl - IRC Chatlog
Search
13:56:36
scymtym
Bike: sorry for bringing this up once again but did you get a chance to test the updated eclector changes?
13:57:28
Bike
ah, no, sorry, yesterday ended up being a bit hectic. i can do it today. i'll just get to a stopping point with what i'm doing first
14:11:52
beach
scymtym: Am I right that eclector.parse-result:make-source-range is deprecated, yet called by Eclector code, so that a style warning is issued, when Eclector is compiled?
14:19:08
Bike
without further changes to clasp from when i updated it for the earlier sharpsign-equals-protocol, i get an error where something wants a cst but gets nil. curious...
15:00:01
scymtym
beach: i renamed the source position and range functions so that their names are now in the ECLECTOR.BASE package (in order to be able to make source ranges for error conditions without circularly depending on the parse-result module). for backward compatibility, i added functions which use the old names
15:00:12
scymtym
the "delegation" logic includes a call from the "new" functions to the "old" functions. compiling this call generates the deprecation warning. i plan to remove the backward compatibility code some time in the future which should resolve the warning
15:02:13
scymtym
Bike: it is possible that, compared to the previous integration attempt, you have to adapt the logic for deciding whether an object is a finalized labeled object, a non-finalized labeled object or an ordinary object
15:05:39
Bike
wow, i think fixup-place has the most complicated lambda list i've ever seen in real code.
15:07:28
Bike
i don't understand what the circular-labeled-object-forms are for, so hopefully i can just leave that empty. the problem might just be that the fixup method didn't know about final/circular
15:07:56
scymtym
sorry. it was supposed to be like CASE but maybe the differences so substantial that it just ends up confusing
15:08:55
scymtym
you can likely ignore that case. at the moment, that case is used only for fixing up hash tables
15:23:45
Bike
just wondering if the clasp-specific structures i'm fixing up are similarly complicated
15:24:18
scymtym
something like (let ((eclector.reader:*readtable* *hashtable-readtable*)) (eclector.reader:read-from-string "#1=(#2=#{#1# (b #2#)})")) when FIXUP is called for #2= before the FIXUP call for #1=. this is hard to write with lisp syntax but there is a test case that constructs an object graph which triggers the case
15:25:54
beach
scymtym: Sure, but it seems the test directory refers directly to the deprecated function.