nicktick has quit [Read error: Connection reset by peer]
<luis>
I wonder how REPL presentations stick around when *object-to-presentation-id* is a weak key hash-table and *presentation-id-to-object* is a weak value hash-table.
<luis>
I don't see return values being stored anywhere else, except for * ** ** /// // /
<luis>
oh, and in fact they don't stick around.
<luis>
I'd never noticed this. Trying to decide whether it's a bug or a feature.
<luis>
Everyone (2 people) in #lisp seems surprised about presentations weakly referencing their respective objects, FWIW.
<adlai>
of course, because #lisp is only weakly linked to people who actually use the damn thing
<adlai>
one could, with some degree of arbitraritude, declare that CL:INTERN processes strings consisting solely of one repeated character from the subset of operation designators that includes #\* and #\/ to be one of the few and only weak reference hints in the common lisp standard, although that'd be slightly past the ANSI
<adlai>
e.g., dumping garbage in the repl and expecting the garbage collector to live past even one invocation is nasal spark levels of undefined behavior
<adlai>
so, best to cap around some reasonably small fixnum, e.g. array-dimension-limit
<adlai>
luis: am I needed to take this thread back to #lusp?
<luis>
adlai: nah, that's OK, we can discuss here. I'm not sure about capping. Are you being ironic about array-dimension-limit being a small fixnum? It's 4611686018427387903 on SBCL.
<luis>
Certain limits will be reached if you eval 4611686018427387903 expressions in the slime-repl. ;-)
* adlai
aimed between array-total-size-limit and the number of symbols plausibly found in the COMMON-LISP-USER package
cage_ has joined #slime
<adlai>
luis: it's been long enough since I last read any of the slime code, so please include a package designator at least once in a message naming symbols that could be either emacs-side, or in one of the swank backends
<adlai>
e.g. (apropos-list "presentation-id" ())
<adlai>
perhaps it could be useful to poll what's considered a typical *slime-repl c-luser* character mass?
<adlai>
that being the upper bound for mutually exclusive presentations.
<adlai>
the bound can be hardened by hypothesizing that SLDB and inspector interaction is mediated via comments printed to the repl
* adlai
clarifies: :|*slime-repl c-luser*| being the emacs buffer containing the poor idiot trapped outside of the lisp image.
_whitelogger has joined #slime
cage_ has quit [Quit: Leaving]
scymtym has quit [Remote host closed the connection]