Sorry, Henry - I just checked a dyadic function with jpm - STILL ALL
ZEROS IN/UP TO J903.
HOWEVER - it is indeed reporting numbers of entries and times by line in
J904, for the dyad as for monads !
So - Thanks for that, and sorry that I didn't check. I'd assumed that
you'd been too
busy to deal with this issue for this beta release.
Cheers,
Mike
On 12/04/2022 16:45, Henry Rich wrote:
I thought I fixed that a while back & I believe I got good results in
JPM. I'll look again next time I have to profile.
Henry Rich
On 4/11/2022 1:03 PM, 'Michael Day' via Programming wrote:
Thanks.
BTW, not a Beta problem, I realise, and I don't know how many
users are affected, but
I'm still having to make functions monadic if I want showdetail to
provide a useful report;
dyads still yield zero times for all lines - last time I looked,
anyway! po
Mike
On 11/04/2022 17:57, Henry Rich wrote:
Ah, that's telling us showdetail should be updated. Thanks for the
report.
Henry Rich
On 4/11/2022 12:53 PM, 'Michael Day' via Beta wrote:
Is this an example of the error messages/warnings enabled with
9!:55 ] 1e6;' ?
commas i.10 NB. no problem reported....
0,1,2,3,4,5,6,7,8,9,
But suppose I wish to examine the performance of commas(!!):
start_jpm_ 1000000000
17857142
timex'commas i.10'
4.11e_5
showdetail_jpm_ 'commas'
recorded all lines
used and max record count: 24 17,857,142
(004) m}"n should be executed without " using leading axes of a: in m
(004) m}"n should be executed without " using leading axes of a: in m
(004) m}"n should be executed without " using leading axes of a: in m
Time (seconds)
+--------+--------+---+-----------------------------------+
|all |here |rep|commas |
+--------+--------+---+-----------------------------------+
|0.000030|0.000030|1 |monad 1 |. ' ' -.~ [: , ','&,@:":"0|
+--------+--------+---+-----------------------------------+
The three diagnostics (004) m"}n ... are evidently evoked by
something in
or called by showdetail_jpm_ .
I discovered this just now, examining the performance of something a
bit more interesting than "commas", but decided this example
demonstrates
the point of interest.
Cheers,
Mike
On 09/04/2022 20:03, Henry Rich wrote:
It has long been known that J will execute many meaningless
sentences without warning. Examples are
NB **********************************
#if defined(AVX)
perform_important_function misspelled_noun
each of which does nothing.
It has been difficult to give warnings only on true errors. J904
implements a warning, based on the observation that the erroneous
sentences produces verbs, and taking into account contexts where a
verb result would be valid.
This warning occasioned an unusually wide range of opinion at J
Headquarters. Is it: tantamount to a Critical Security Patch,
fixing a gaping hole in the language; a novelty, perhaps of
interest to a few; an unwarranted usurpation of a programmer's
prerogatives, and likely to break working code?
Thus the warning is optional in beta-a. We would like to get good
data on whether the warning is welcome, noxious, or intolerable. I
ask all beta users to enable the warning by adding
9!:55 ] 1e6;''
to your startup script. After a few days, report your findings in
this Forum. Make note of any valid lines that were flagged, and
also of any cases where the warning revealed code to be meaningless.
Henry Rich
--
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
