[Bug-glpk] [Fwd: Re: glpk prints warnings which lead to failing sagemath tests]

[Andrew Makhorin]

-------- Forwarded Message --------
From: Tobias Hansen <address@hidden>
To: Andrew Makhorin <address@hidden>
Cc: Sébastien Villemot <address@hidden>, address@hidden,
address@hidden
Subject: Re: glpk prints warnings which lead to failing sagemath tests
Date: Wed, 28 Feb 2018 11:42:20 +0100

On 02/28/2018 09:51 AM, Andrew Makhorin wrote:
> On Mon, 2018-02-26 at 18:56 +0100, Tobias Hansen wrote:
>> On 02/26/2018 04:46 PM, Andrew Makhorin wrote:
>>> On Mon, 2018-02-26 at 12:00 +0100, Sébastien Villemot wrote:
>>>> Dear GLPK maintainer,
>>>>
>>>> I received the attached bug report and patch for the glpk package in 
>>>> Debian.
>>>>
>>>> What’s your take on this? Should I apply the patch, or is this something 
>>>> that
>>>> should better be fixed at the sagemath level?
>>>>
>>>> Best,
>>>>
>>>> P.S.: please keep all addresses in CC when replying.
>>> Thank you for your bug report.
>>>
>>> Could you please explain in more details how glpk is tested? Both glpk
>>> lp and mip solvers issue many messages to the standard output, so it is
>>> unclear to me how some of the messages may affect the tests.
>>>
>>> Andrew Makhorin
>> Hi,
>>
>> one example that I checked was using the mip solver (not sure if other
>> solvers are affected) with msg_lev set to GLP_MSG_OFF. Normally there
>> is no terminal output, now there is the message "Long-step dual
>> simplex will be used". GLPK is used from a Python program which checks
>> the terminal output in its test suite.
> GLP_MSG_OFF is not a relevant option to disable terminal/stdout output. 
> Please use glp_term_out(GLP_OFF) and glp_term_out(GLP_ON) instead--these
> calls disable/enable terminal output on a lower level independently on
> options passed to the solvers.
>
> Besides, as was noticed by Heinrich, it is not a good idea to parse the
> terminal output to test glpk (what namely do you want to test in such a
> way?).
>
> Best regards,
>
> Andrew Makhorin

The terminal output that is parsed is the output of the program that is linked 
against glpk. And it seems Octave also gets these messages in its output now, 
see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=891465#20

Best,
Tobias


>
>
>>>> MHTML Document attachment (Bug#891465: glpk: prints warnings which
>>>> lead to failing sagemath tests)
>>>>> -------- Forwarded Message --------
>>>>> From: Tobias Hansen <address@hidden>
>>>>> Reply-to: Tobias Hansen <address@hidden>, address@hidden
>>>>> To: address@hidden
>>>>> Subject: Bug#891465: glpk: prints warnings which lead to failing
>>>>> sagemath tests
>>>>> Date: Sun, 25 Feb 2018 20:54:19 +0100
>>>>>
>>>>> Source: glpk
>>>>> Version: 4.65-1
>>>>> Severity: normal
>>>>> Tags: patch
>>>>>
>>>>> Hi there,
>>>>>
>>>>> since version 4.65 glpk started to frequently print the message
>> "Long-step dual simplex will be used", leading to many failed tests
>> for sagemath and sagemath failing to build. Not sure if it's a bug in
>> glpk but it seems to me it shouldn't print this. Could you maybe
>> deactivate the messages to allow sagemath to build?
>>>>> Best,
>>>>> Tobias
>>
>>
>