[Rivet] Wrong latex configuration on some systems

[Andy Buckley]

On 17/03/14 13:54, Oldrich Kepka wrote:
> Hi Andy,
> 
> On 17 March 2014 13:27, Andy Buckley <andy.buckley at cern.ch> wrote:
>> Ok, thanks Frank. I am slightly wary since this has been such an
>> unpredictable area before, but since it's a change of less than a factor
>> of 10 we probably didn't explore this value yet.
>>
>> Oldrich, how did you reach a value of 12M? Was that the highest round
>> value that didn't crash your system, or is it a bit more ad hoc than
>> that? e.g. would 15M also work?
> 
> I took it from the troubleshooting page linked above. The highest
> acceptable working limit is:
> 12435455

Aha, I didn't notice that was in the same units as the config file. In
that case 12M certainly makes sense as a default. Funny that we got away
with 70M on so many systems!

Andy


>> I'll certainly let you know if it messes up on my machine before we make
>> another release ;-)
>>
>> In the long run I think re-writing make-plots is still a very good idea,
>> although of course none of us get much academic cachet for doing such
>> things. I recently started again on some hacking with David Mallows'
>> amazing TeX/Cairo hybrid rendering system, after a 6 month break --
>> playing with plot cosmetics is rather fun since you see the results
>> immediately! It looks very much like make-plots, but is technically a
>> lot nicer and much faster. I'll circulate a test version to this Rivet
>> dev list for people to play with, once I've made it a bit more
>> presentable, and hopefully we can transition to that in the next year
>> (I'm being realistic: it will take that long to find manpower to
>> translate the make-plots logic and debug to the extent that's needed).
>>
>> Andy
>>
>>
>> On 17/03/14 13:20, Frank Siegert wrote:
>>> Hi Andy, Oldrich,
>>>
>>> I think going from 70M to 12M should be acceptable, if that gives us
>>> an additional system where it works out of the box. I have committed
>>> this now and pushed it into the release branch.
>>>
>>> Cheers,
>>> Frank
>>>
>>> On 17 March 2014 10:02, Andy Buckley <andy.buckley at cern.ch> wrote:
>>>> I thought it already was -- thanks for the reminder! Now done.
>>>>
>>>> Cheers,
>>>> Andy
>>>>
>>>>
>>>> On 17/03/14 09:33, Oldrich Kepka wrote:
>>>>> Hi Andy,
>>>>>
>>>>> understood. What about mentioning this issue at Troubleshooting page
>>>>> until it is resolved (if ever).
>>>>>
>>>>>
>>>>> On 17 March 2014 09:28, Andy Buckley <andy.buckley at cern.ch> wrote:
>>>>>> Hi Oldrich,
>>>>>>
>>>>>> It's a really annoying problem: too small a memory setting and some
>>>>>> plots can't be built; too much and TeX dies with this error. And the max
>>>>>> value is apparently system dependent -- I can guarantee that every Rivet
>>>>>> release has run correctly on my Ubuntu system (including my current 64
>>>>>> bit 13.10) but we do periodically get reports like this.
>>>>>>
>>>>>> Perhaps 32 vs. 64 bit systems matter as well as the OS, but we don't
>>>>>
>>>>> Yes, that might be, I have 32 bit.
>>>>>
>>>>> Cheers,
>>>>> Oldrich
>>>>>
>>>>>
>>>>>> have a list of acceptable values for different systems, or any idea of
>>>>>> how to test for it... other than trying out different values in the
>>>>>> configure script and counting down until TeX doesn't crash. I don't have
>>>>>> time to try writing and integrating something like that right now (and
>>>>>> when I do I'd rather spend it on writing a more modern replacement for
>>>>>> make-plots with fewer TeX requirements!) but maybe someone else would
>>>>>> like to take up that challenge.
>>>>>>
>>>>>> Andy
>>>>>>
>>>>>>
>>>>>> On 17/03/14 09:07, Oldrich Kepka wrote:
>>>>>>> Hi all,
>>>>>>>
>>>>>>> installing 2.1.0 I get the following error
>>>>>>>
>>>>>>> olda at marlowe:~/work/atlas/moudra/ZZ_ATLAS_2012_I1203852$ rivet-mkhtml
>>>>>>> Rivet.yoda
>>>>>>> Making 13 plots
>>>>>>> Plotting ./plots/ATLAS_2012_I1203852/ZZs_ZpT.dat (12 remaining)
>>>>>>> Plotting ./plots/ATLAS_2012_I1203852/ZZs_mZZ.dat (11 remaining)
>>>>>>> dvips: DVI file can't be opened: ZZs_ZpT: No such file or directory
>>>>>>>
>>>>>>> which with -v option resolves to
>>>>>>>
>>>>>>> dvips d06-x01-y01 -f
>>>>>>> Ouch---my internal constants have been clobbered!---case 14
>>>>>>>
>>>>>>> and is a known issue:
>>>>>>>
>>>>>>> http://tex.stackexchange.com/questions/67014/latex-gives-me-the-error-ouch-my-internal-constants-have-been-clobbered
>>>>>>>
>>>>>>>
>>>>>>> Chaning: main_memory = 70000000 to main_memory = 12000000 in
>>>>>>>
>>>>>>> in share/Rivet/texmf/cnf/texmf.cnf
>>>>>>>
>>>>>>> solves the problem. It would be nice to have the main_memory set to
>>>>>>> appropriate values depending on tex instalation distribution, though I
>>>>>>> have not looked into that myself.  I am using generic ubuntu 13.10
>>>>>>> texlive-bibtex-extra installation.
>>>>>>>
>>>>>>> Cheers,
>>>>>>> Oldrich
>>>>>>> _______________________________________________
>>>>>>> Rivet mailing list
>>>>>>> Rivet at projects.hepforge.org
>>>>>>> http://www.hepforge.org/lists/listinfo/rivet
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Dr Andy Buckley, Royal Society University Research Fellow
>>>>>> Particle Physics Expt Group, University of Glasgow / PH Dept, CERN
>>>>
>>>>
>>>> --
>>>> Dr Andy Buckley, Royal Society University Research Fellow
>>>> Particle Physics Expt Group, University of Glasgow / PH Dept, CERN
>>>> _______________________________________________
>>>> Rivet mailing list
>>>> Rivet at projects.hepforge.org
>>>> http://www.hepforge.org/lists/listinfo/rivet
>>
>>
>> --
>> Dr Andy Buckley, Royal Society University Research Fellow
>> Particle Physics Expt Group, University of Glasgow / PH Dept, CERN


-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN