Hi All,
I've asked author of profphd changelog entry for details why profphd relies
on older version of perl and perlbrew. His answer is given below.
I'll try to fix profphd now - last time I tried to launch it with latest
perl it there was no such error.

---------- Forwarded message ----------
From: Timothy Karl <[email protected]>
Date: 2016-07-18 16:18 GMT+03:00
Subject: Re: on Debian "profphd' upstream ChangeLog's entry
To: merlettaia <[email protected]>


Hi Tanya,

I checked back in my logs and found the following that I wrote on March 20,
2015, before the official release of Jessie:

It appears that at line 7739 in /usr/share/profphd/prof/scr/lib/lib-prof.pl:

   $tmpend[$cth]=$tmplen[$cth] if (! defined $tmpend[$cth]);

The command that is being run is:
   prof query.hsspPsiFil both fileRdb=query.profRdb numresMin=17
nresPerLineAli=60 riSubSec=4 riSubAcc=3 riSubSym=.

Both $tmpend[$cth] and $tmpend[$cth] are undefined at this point on the
first iteration of the function. On the cluster node (using perl 5.14), the
assignment works; however, on debian-unstable (using perl 5.20), prof fails
with "Modification of non-creatable array value attempted, subscript -1 at
/usr/share/profphd/prof/scr/lib/lib-prof.pl line 7739.”, although at this
point on both servers the value of $cth is 0.


At the time, prof just failed using perl 5.20 installed on debian-unstable
(now jessie), giving the aforementioned error message. Arybase wasn’t, at
the time, implementing the ‘$[‘ assignment the same as in previous versions
of Perl core. The developer or prof, who created it years ago, did not like
counting from 0 on arrays, and therefore, used the ‘$[' to start array
counting at 1 (instead of 0). Since I was preparing for the release of
Jessie, and since there are literally thousands upon thousands of poorly
documented perl code in that package, of which I’m not the maintainer nor
contributor, I decided it was easiest and best to use perlbrew, including
it as a requirement, and having it install an older version of perl with
which prof could run against.

Recently, a college had tried the previous version of prof, before the
perlbrew became a requirement, on jessie. He said it didn’t produce an
error any longer. Perhaps arybase has been changed, making it that perlbrew
is no longer needed for prof to run correctly.

So basically, if you don’t get the error message, then prof ran as it
should.  I hope this helps you :)
Cheers,
Tim

Email : [email protected]
Jabber: [email protected]

>>Encrypted emails are preferred<<




On 17 Jul 2016, at 19:51, merlettaia <[email protected]> wrote:

Hi Timothy,
I'm an intern in Debian Med Team, and I'm working on autopkgtest testsuites
for various packages. I found your email in profphd upstream's ChangeLog.

"
profphd (1.0.42) unstable; urgency=low

  * Make prof work with Debian Jessie by requiring perlbrew package
    and have it install independent perl v5.10 instance. $[ has been
    moved from perl core >=5.16.0 into arybase, and does not produce
    the same side effects as in previous perl versions, therefore
    breaking prof when run using perl v5.20 in Jessie.

 -- Timothy Karl (Aerialis) <[email protected]>  Thu, 23 Jul 2015 16:44:30
+0200
"

Can you remember what exact error was produced? And what erratic behaviour
did you noticed?
I've read arybase description here http://search.cpan.org/dist/pe
rl-5.17.1/ext/arybase/arybase.pm#HISTORY, and for me it is unclear what
specific side effect makes reprof break in latest versions of perl (and how
exactly this breaks reprof?).

Unfortunately, I don't speak Perl fluently now and can't afford myself full
source code investigation - my internship ends in August, and I wish to
write tests for as many other packages as possible before it ends (and I've
already spent 1 month while working on RostLab's packages). Also, I
couldn't make perl 10.1 work in autopkgtest testsuite for now, even with
perlbrew installed, so I'll try to make patch to update package to use
latest Perl version if it is possible.
That's why any information would be helpful.

Thank you for any answer.

-- 
Regards,
Tanya.





-- 
Best wishes,
Tanya.

Attachment: smime.p7s
Description: S/MIME cryptographic signature

Reply via email to