jibanes has quit [Ping timeout: 256 seconds]
jibanes has joined #picolisp
GrayArea has quit [Quit: leaving]
orivej has quit [Ping timeout: 256 seconds]
jibanes has quit [Read error: Connection reset by peer]
jibanes has joined #picolisp
<
Regenaxer>
I was wrong yesterday. The alignment in pil21 was all correct
<
Regenaxer>
The printed numbers for 'A'
*are* aligned
<
Regenaxer>
: (hex 549612477024)
<
Regenaxer>
-> "7FF774DA60"
<
Regenaxer>
So the change yesterday was unnecessary
<
Regenaxer>
The default alignment is 8
<
Regenaxer>
The change is not wrong though, it just confirms the default
<
Regenaxer>
So the problem remains what exactly goes wrong in llvm12
<
Regenaxer>
I reverted src/lib/llvm.l
<
Regenaxer>
(really should not start such stuff in the evening ;)
_whitelogger has joined #picolisp
<
Regenaxer>
good morning tankf33der!
<
tankf33der>
installing virtualbox and fedora rawhide to get llvm12 on laptop without work access everytime
<
Regenaxer>
Perhaps search backwards, where the data are wrong first?
<
tankf33der>
will do asap
<
Regenaxer>
we knew that (car P) was wrong
<
Regenaxer>
the above shows if 'eval' already returns the wrong value
<
Regenaxer>
'L' was all right
<
Regenaxer>
strange!
orivej has joined #picolisp
<
tankf33der>
i head is bad, ive setup wrong os
<
tankf33der>
i cant explain how it happened
<
tankf33der>
i must slowdown
<
Regenaxer>
No hurry
<
tankf33der>
Regenaxer: should i replace this _stem function?
<
tankf33der>
or just add several lines?
<
tankf33der>
i got ubuntu access with llvm12
<
Regenaxer>
"this" stem function is the one I posted?
<
Regenaxer>
It is an example
<
Regenaxer>
You could experiment with it
<
Regenaxer>
It outputs the unevaluated arguments with (dbg 1 X)
<
Regenaxer>
and the the result of evaluation with (dbg 2 (eval (car X)))
<
Regenaxer>
The idea is to see where it is still correct and which operation gives the wrong result
<
Regenaxer>
We saw (NIL . @) in your post
<
Regenaxer>
Perhaps you can also test simpler calls
<
Regenaxer>
(stem (1 2 3) 2) or so
<
Regenaxer>
easier to track down the problem
<
tankf33der>
after hours
<
Regenaxer>
I'm also away till early afternoon
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #picolisp
<
tankf33der>
$ ./pil
<
tankf33der>
: (stem (1 2 3) 2)
<
tankf33der>
-> (1 2 3)
<
tankf33der>
llvm12
<
tankf33der>
-> (3)
<
tankf33der>
../pil
<
tankf33der>
: (stem (1 2 3) 2)
<
tankf33der>
llvm11
<
Regenaxer>
OK, so 'eval' works, but 'equal' fails
<
Regenaxer>
What is printed for the original long argument list?
<
tankf33der>
: (stem (chop "abc/def\\ghi") "/" "\\")
<
tankf33der>
$ ./pil
<
tankf33der>
2 "\\"
<
tankf33der>
1 ("/" "\\")
<
tankf33der>
-> ("a" "b" "c" "/" "d" "e" "f" "\\" "g" "h" "i")
<
tankf33der>
llvm12
<
Regenaxer>
but different from yesterday
<
Regenaxer>
Can you try the one from yesterday again? Printing (car P)?
<
tankf33der>
i need time to find code
<
Regenaxer>
I suggest (dbg 3 (car P)) just before 'equal'
<
Regenaxer>
this was the place yesterday iirc
<
tankf33der>
trying
<
tankf33der>
../pilmpech@ubuntu:~/pil21/src$ ../pil
<
tankf33der>
mpech@ubuntu:~/pil21/src$ ../pil
<
tankf33der>
mpech@ubuntu:~/pil21/src$
<
tankf33der>
: (stem (1 2 3) 2)
<
tankf33der>
Segmentation fault (core dumped)
<
tankf33der>
: (stem (chop "abc/def\\ghi") "/" "\\")
<
tankf33der>
1 ("/" "\\")
<
tankf33der>
2 "\\"
<
tankf33der>
Segmentation fault (core dumped)
<
tankf33der>
: (stem (1 2 3) 2)
<
tankf33der>
================= ../pil
<
tankf33der>
-> (3)
<
tankf33der>
this is llvm12 and works now
<
tankf33der>
removed -O3 from opt
<
tankf33der>
this is what i seen a year ago
<
tankf33der>
when pil21 crashed on opt -O3
<
tankf33der>
summary
<
tankf33der>
i believe now this is not pil21 issue!
<
Regenaxer>
mom, I'm back home soon
<
Regenaxer>
I see. So a bug in the optimizer?
<
tankf33der>
seems so
<
tankf33der>
we must be warned to use stable only versions of LLVM then
<
tankf33der>
i dont know how you broke it and how they find and fix it back
<
Regenaxer>
Good, so let's just wait
<
Regenaxer>
till 12 is stable
<
tankf33der>
i could write a message to llvm-dev about issue and offer a help how to test llvm by pil21 :)
<
Regenaxer>
I'm afraid it is too complicated for them to set up and track down the error
<
tankf33der>
just use pil21 as marking flag.
<
Regenaxer>
marking flag?
<
tankf33der>
to use pil21 as flag if llvm is production ready :)
<
Regenaxer>
yes, right
<
Regenaxer>
the ultimate test ;)
<
Regenaxer>
pass manager, optim pipeline
<
Regenaxer>
sounds complicated
<
tankf33der>
lets see.
<
Regenaxer>
Final version (from yesterday) is now in Debian unstable
<
Regenaxer>
Tested, looks all good
<
Regenaxer>
e.g. also bash completion
orivej has quit [Ping timeout: 265 seconds]
mmmattyx has joined #picolisp
Blukunfando has joined #picolisp
orivej has joined #picolisp
<
rick42>
wow, you guys! great work! tankf33der, i hope your shoulder is feeling well now
<
Regenaxer>
Hi rick42! :)
<
rick42>
hey Regenaxer! :D
<
rick42>
so glad to see you all working this out well.
<
Regenaxer>
Let's hope so ;)
<
beneroth>
great job
<
Regenaxer>
Hi beneroth, thx
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #picolisp
mmmattyx has quit [Quit: Connection closed for inactivity]