These are chat archives for dropbox/pyston

26th
Aug 2015
Sun
@Daetalus
Aug 26 2015 01:08
@kmod #817 updated and waiting for review. Would you mind reivew it if you have time?
Kevin Modzelewski
@kmod
Aug 26 2015 01:45
ok :)
Marius Wachtler
@undingen
Aug 26 2015 10:34
@kmod can you please take a look at Box::appendNewHCAttr. I think the rewriter path of this functions is very strange. Should it really always reuse a constant as size argument to the gc_realloc call etc...?
Kevin Modzelewski
@kmod
Aug 26 2015 19:59
I think that should be ok?
well, maybe not for the SINGLETON case
Kevin Modzelewski
@kmod
Aug 26 2015 20:18
I think ASTInterpreter::doOSR is the place we lose type information when osr'ing
we just pass all variables as UNKNOWN
looks like I didn't push my experiment of passing xrange iterators as typed;
I think the issue was that we would have to make sure we don't reuse an OSR exit with the wrong type signature
(right now we are guaranteed they will always have the same signature so I don't think it checks)
we might want though to adjust the tiering strategy/constants so that we don't take the OSRs more than once
Kevin Modzelewski
@kmod
Aug 26 2015 20:41
ooh just found the info break command in gdb
shows you how many times a breakpoint was hit
I kept on adding a counter to the code to make sure I could breakpoint it at the right time
Marius Wachtler
@undingen
Aug 26 2015 20:44
:-) didn't know that command