Behalf Of
Gilles Kirouac
Sent: Tuesday, August 7, 2018 11:25 AM
To: programm...@jsoftware.com
Subject: Re: [Jprogramming] g420fg NaN error (defining a dyadic verb)
To define alpha1 as a dyadic verb, use
alpha1=: 4 :'y /: <./x i."1 _ y' NB. 4 instead of 3
~ GIlles
Le 2018-08
Curses. Thanks for checking.
Henry
On 8/7/2018 12:11 PM, Bill Heagy wrote:
Doesn't help, even after changing FE_ALL_EXCEPTS to FE_ALL_EXCEPT (no
"S").
On 2018-08-07 11:34 AM, Henry Rich wrote:
I pushed something else for you to try. clang reportedly supports
feclearexcept from the C standard
Doesn't help, even after changing FE_ALL_EXCEPTS to FE_ALL_EXCEPT (no "S").
On 2018-08-07 11:34 AM, Henry Rich wrote:
I pushed something else for you to try. clang reportedly supports
feclearexcept from the C standard. Maybe it supports it better than
_clearfp. I have pushed a version that us
I pushed something else for you to try. clang reportedly supports
feclearexcept from the C standard. Maybe it supports it better than
_clearfp. I have pushed a version that uses feclearexcept if the symbol
__STDC_IEC_559__ is defined. I read rumors than clang sets this symbol,
but I can't c
To define alpha1 as a dyadic verb, use
alpha1=: 4 :'y /: <./x i."1 _ y' NB. 4 instead of 3
~ GIlles
Le 2018-08-07 à 03:46, Linda Alvord a écrit :
I can't seem to get messages to the forum. Here's a problem in J806
a=:'abcdefghijklmnopqrstuvwxyz'
A=:'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
aA=:
s
|[-19] c:\users\linda\j64-806-user\temp\32.ijs
Linda
-Original Message-
From: Programming On Behalf Of Henry
Rich
Sent: Monday, August 6, 2018 11:01 PM
To: programm...@jsoftware.com
Subject: Re: [Jprogramming] g420fg NaN error
Thanks. I guess we have to figure out how to convince
Ok, I won't expect that tonight B-)
On 2018-08-06 11:20 PM, Henry Rich wrote:
I have reverted the latest push and await a compiler fix.
Henry Rich
On 8/6/2018 11:01 PM, Bill Heagy wrote:
SVN 861 for the gcc/clang tests and the 2-parts test
On 2018-08-06 10:59 PM, Henry Rich wrote:
Did you c
I have reverted the latest push and await a compiler fix.
Henry Rich
On 8/6/2018 11:01 PM, Bill Heagy wrote:
SVN 861 for the gcc/clang tests and the 2-parts test
On 2018-08-06 10:59 PM, Henry Rich wrote:
Did you check out the latest push to see if that makes any difference?
Henry Rich
On 8/
Thanks. I guess we have to figure out how to convince clang to create
the right code then, eh?
Henry Rich
On 8/6/2018 10:59 PM, Bill Heagy wrote:
And yes, both parts fail (assuming you're talking about the failing
line in g420fg)
On 2018-08-06 10:50 PM, Bill Heagy wrote:
clang with -O0 fix
SVN 861 for the gcc/clang tests and the 2-parts test
On 2018-08-06 10:59 PM, Henry Rich wrote:
Did you check out the latest push to see if that makes any difference?
Henry Rich
On 8/6/2018 10:50 PM, Bill Heagy wrote:
clang with -O0 fixes it too.
On 2018-08-06 10:44 PM, Bill Heagy wrote:
Bui
And yes, both parts fail (assuming you're talking about the failing line
in g420fg)
On 2018-08-06 10:50 PM, Bill Heagy wrote:
clang with -O0 fixes it too.
On 2018-08-06 10:44 PM, Bill Heagy wrote:
Building with gcc works ok. I'll try clang/-O0
On 2018-08-06 10:01 PM, bill lam wrote:
Just a
Did you check out the latest push to see if that makes any difference?
Henry Rich
On 8/6/2018 10:50 PM, Bill Heagy wrote:
clang with -O0 fixes it too.
On 2018-08-06 10:44 PM, Bill Heagy wrote:
Building with gcc works ok. I'll try clang/-O0
On 2018-08-06 10:01 PM, bill lam wrote:
Just a gues
clang with -O0 fixes it too.
On 2018-08-06 10:44 PM, Bill Heagy wrote:
Building with gcc works ok. I'll try clang/-O0
On 2018-08-06 10:01 PM, bill lam wrote:
Just a guess, recently we switched from gcc to clang. clang is too
smart in
optimization that it short circuit some code that deals wit
Building with gcc works ok. I'll try clang/-O0
On 2018-08-06 10:01 PM, bill lam wrote:
Just a guess, recently we switched from gcc to clang. clang is too smart in
optimization that it short circuit some code that deals with NaN, I've put
some inline asm to teach clang register flags need checki
Just a guess, recently we switched from gcc to clang. clang is too smart in
optimization that it short circuit some code that deals with NaN, I've put
some inline asm to teach clang register flags need checking. This works in
the past.
Can you try build and test with -O0 or with gcc?
On Tue, Aug
This baffles me. I wonder if there is some compiler interaction with
_clearfp(). In case so, I have pushed out another change that puts a
couple of memory fetches between the clearfp()s. Would you please build
that & report back? If it fails on the same line, please try the two
parts individ
If it can be of any help, everything is fine in my older J version in Linux
:
etx =: 1 : 'x :: (<:@(13!:11)@i.@0: >@{ 9!:8@i.@0:)' NB. error message
'domain error' -: +/@:*"1~ etx u: 'abc'
1
0 _ 1 (+/@:*"1 -: +/@:*"_"1) _ 0 2 NB. removable NaN
1
JVERSION
Engine: j701/2011-01-10/11:25
Librar
Can you help me debug this? It works on Windows and I don't see any
problem in the code.
The sections of interest are at va2.c:787 and va2.c:840. They run the
two verbs, either or both of which might be failing. The earlier is for
+/@:*"1 .
This is line 787:
NAN0;
D *av=DAV(a),*wv=D
Ah...
In that case, etx is from
https://github.com/jsoftware/jsource/blob/master/test/tsu.ijs#L62
etx =: 1 : 'x :: (<:@(13!:11)@i.@0: >@{ 9!:8@i.@0:)' NB. error message
from error number
So, basically, the test is whether
+/@:*"1~ u: 'abc'
produces a domain error.
And, since it does produc
Clarifying, I'm running "RUN1 'g420fg'" or equivalent, and just quoting
the end output.
On 2018-08-06 06:28 PM, Raul Miller wrote:
I get
'domain error' -: +/@:*"1~ etx u: 'abc'
|value error: etx
FYI,
--
For information
I get
'domain error' -: +/@:*"1~ etx u: 'abc'
|value error: etx
FYI,
--
Raul
On Mon, Aug 6, 2018 at 6:26 PM Bill Heagy wrote:
>
> Sorry, I was a bit hasty there. I see it on 64-bit linux too.
>
> On 2018-08-06 06:13 PM, Bill Heagy wrote:
> > 32-bit again?
> >
> > ..
> > 1
> > 'do
Sorry, I was a bit hasty there. I see it on 64-bit linux too.
On 2018-08-06 06:13 PM, Bill Heagy wrote:
32-bit again?
..
1
'domain error' -: +/@:*"1~ etx u: 'abc'
1
NB. Test for NaN
0 _ 1 (+/@:*"1 -: +/@:*"_"1) _ 0 2 NB. removable NaN
|NaN error
| 0 _ 1 (+/@:*"1-:+/@:*"
32-bit again?
..
1
'domain error' -: +/@:*"1~ etx u: 'abc'
1
NB. Test for NaN
0 _ 1 (+/@:*"1 -: +/@:*"_"1) _ 0 2 NB. removable NaN
|NaN error
| 0 _ 1(+/@:*"1-:+/@:*"_"1)_ 0 2
|[-189] /home/wheagy/git/jsource/test/g420fg.ijs
JVERSION
Engine: j806/j32/linux
Beta: GPL3/201
23 matches
Mail list logo
