<Regenaxer>
tankf33der, this message from valgrind "param poll(ufds.events) points to uninitialised byte(s)" is irrelevant
<Regenaxer>
:(
<Regenaxer>
I've spent hours yesterday to find the reason
<Regenaxer>
But as I said these entries are ignored
<tankf33der>
then you should leave it as is.
<Regenaxer>
yeah
<Regenaxer>
But I still have a segfault in GUI
<Regenaxer>
I thought the valgrind message points to the reason
<tankf33der>
segfault is bad.
<Regenaxer>
very true
<Regenaxer>
I reverted the Wiki to pil64 for now
<tankf33der>
do you have coredump file from crash?
<Regenaxer>
No longer. Will try to reproduce
<tankf33der>
dont forget add -g to clang in makefile
<Regenaxer>
iirc it crashed in (accept)
<Regenaxer>
found via -traceAll
<tankf33der>
i will run tiny unless-fork file in gc+
<Regenaxer>
Probably no gc+ issue
<Regenaxer>
and not so easy to reproduce
<tankf33der>
is this crash relevant to huge io changes recently?
<Regenaxer>
I see it only in one app here
<Regenaxer>
but also in the Wiki
<tankf33der>
maybe you forgot to change something internally
<Regenaxer>
I'm not sure if it is new
<Regenaxer>
yeah
<tankf33der>
i would be wrong variable name on stack in some branch somewhere. :)
<Regenaxer>
Yes, could be anything
<tankf33der>
fork uless-fork under gc+ ok
rob_w has joined #picolisp
<tankf33der>
valgrind is ok for pil32
orivej has quit [Ping timeout: 256 seconds]
<Regenaxer>
ok
jibanes has quit [Ping timeout: 240 seconds]
jibanes has joined #picolisp
aw- has joined #picolisp
<Regenaxer>
As yesterday, it crashes in malloc() inside (accept)
<Regenaxer>
in 'initInFile' to be exact
<Regenaxer>
'initInFile' is indeed part of the rewrite
jibanes has quit [Ping timeout: 260 seconds]
<Regenaxer>
As malloc() crashes, the problem must be earlier
<Regenaxer>
Besides the warning yesterday, valgrind also complain about "Address 0x4f82694 is 4 bytes inside a block of size 160 alloc'd". Wtf does that mean?
jibanes has joined #picolisp
<Regenaxer>
I know the reason. It is multiply closing and free'ing file structures
<Regenaxer>
Hard to solve
<Regenaxer>
Seems it had been around all the time. Also in pil32 and pil64. Not sure why it never manifested
<Regenaxer>
Need to rewrite the I/O structures once more
<tankf33der>
uh
<tankf33der>
sounds like expaination of segfaults
<Regenaxer>
yes
<Regenaxer>
Complicated cases, like (catch .. (in .. (task (close ...
<Regenaxer>
'in' closes *again*
<Regenaxer>
closing again is OK, but it free()s also here
<Regenaxer>
good, rewrite is not so very big
<tankf33der>
already released?
jibanes has quit [Ping timeout: 246 seconds]
jibanes has joined #picolisp
<Regenaxer>
No
<Regenaxer>
Not even solved yet
<Regenaxer>
thing for a good way
<Regenaxer>
*thinking
rob_w has quit [Quit: Leaving]
orivej has joined #picolisp
orivej has quit [Ping timeout: 260 seconds]
<Regenaxer>
Hmm, confused
<Regenaxer>
Something else must be wrong
<Regenaxer>
GUI hangs in 'accept' now
<Regenaxer>
I think a
<Regenaxer>
I think I'm searching at the completely wrong places. Worsening ... ;)
<Regenaxer>
Or I indeed messed up something in I/O
<Regenaxer>
I give up now. Continue later perhaps
<aw->
hi Regenaxer
<aw->
is there a way to change the prompt of the picolisp repl?
<aw->
instead of :
<aw->
can i do it without recompiling?
<aw->
i think you had it for the ircclient? but the link is 404