bug-apl
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: /⍨


From: Dr . Jürgen Sauermann
Subject: Re: /⍨
Date: Thu, 9 Jun 2022 15:26:09 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.13.0

Hi Hudson,

thanks, see below. SVN 1561.

Best Regards,
Jürgen


On 6/8/22 3:59 PM, hudson@hudsonlacerda.com wrote:
Hi Jürgen,

----- Em 7 de Jun de 2022, em 13:56, Dr. Jürgen Sauermann mail@jürgen-sauermann.de escreveu:

Hi Hudson,

I believe that I fixed the double execution for executable scripts,
Looks like the OS handles executable scripts differently than
non-executable ones. SVN 1560 .
Thanks.

Now, executable scripts need "-f" to be run directly ( ./script.apl ).

   #!/usr/bin/env -S apl -s -f

Otherwise, apl waits for stdin input (the script is not executed),
and apl cannot be terminated by killall apl — only by ")off" or kill <PNUM>.
Fixed. However, please note the following.

The operating system, at least GNU/Linux resp. bash handles text files
differently depending on whether they are executable or not. This is
not detectable by the executed script or binary (= apl). The only thing
that apl can do is to apply some heuristics as to how it was started.

You may (and should)  avoid the above hanging by having )OFF at the
end of the script or by using the --OFF command line option (or in the
shebang line if the script is started directly).

The F ← {⍵ × (?⍨⍵) ,¨ ⍳⍵} bug is something that I cannot reproduce.
I wonder if it happens always or lnly sometimes.
It happened always, for ⍵>35 around. But the errors were different
for every argument value. See attachment.


I have some suggestions
for you and others that make my life easier:

* run " make develop " in the top-level directory.

[...]

Thank you for the explanation. It was useful.
After some testing, my conclusions are:


1) The errors seem be caused by parallel processing. I was using

   ./configure DEVELOP_WANTED=yes CORE_COUNT_WANTED=2

   (Now I run ./configure without options.)

I have change the code so that A?B (which was probably the culprit)
will no longer be executed in parallel.

2) With  make develop / make apl.lines / ./apl, the program worked fine
   inside src/ directory.


3) I got the same bugs after "sudo make install", running /usr/local/bin/apl.


4) No problems if "sudo make uninstall" before reconfiguring and recompiling.
   That suggests me that some residues of previous versions were remaining.


5) README-2-configure suggests running

   autoreconf --force --install
   ./configure
   make

   but I need also calling autoupdate:

   autoupdate
   autoreconf --force --install
   ./configure
   make


Best regards,
Hudson







reply via email to

[Prev in Thread] Current Thread [Next in Thread]