[Users] Range error setting parameter, on cluster

[Roland Haas]

Hello Jordan,

Adding this to the ticket would be great. 

On my workstation I have gcc-15 and things work well there, so this may
be only some specific version that is affected, so knowing the exact
number is helpful (one could even compile that version and test).

Yours,
Roland

> [CAUTION: Non-UBC Email]
> 
> Hi Roland (and Peter),
> 
> 
> Thank you for the suggestion! This was very relevant.
> 
> With loaded modules related to GCC-13.3.0 (and GCC-14.3.0), the error is occurring, while for GCC-12.3.0, the range is working as expected. This is compatible with the experiment on other clusters and on my work station, which have lower versions of gcc as well.
> 
> 
> What should be made of that knowledge? Should I file a ticket? I suppose that in the future, this may be ever more likely to occur. (For my individual case, it's fine, I can deal with it.)
> 
> 
> Best,
> 
> 
> Jordan
> 
> ________________________________
> From: Roland Haas <rhaas at mail.ubc.ca>
> Sent: Thursday, March 12, 2026 2:33:03 PM
> To: Jordan Nicoules via Users
> Cc: Jordan Nicoules
> Subject: Re: [Users] Range error setting parameter, on cluster
> 
> 
> CUIDADO: Email de um sistema externo. Cuidado com links, anexos e pedidos de dados/senhas.
> CAUTION: Email from an external system. Be careful with links, attachments, and requests for data/passwords.
> 
> Hello Jordan,
> 
> We discussed this some more in today's ET call and Peter Diener
> suggested trying a different compiler on Deucalian just in case this is
> actually a compiler issue.
> 
> Yours,
> Roland
> 
> > [CAUTION: Non-UBC Email]
> >
> > Hi Roland,
> >
> >
> > Indeed, Deucalion is a cluster (as well as MesoPSL and MareNostrum, which I checked before sending my previous email, and which seem to work just fine with the "*" range as expected).
> >
> > I'm attaching the requested files:
> >
> > - for the work station: relayer_ubuntu.cfg, workstation_make.config.defn
> >
> > - for the Deucalion cluster: deucalion-x86.cfg, deucalion_make.config.defn
> >
> >
> > These are from the minimal working example I was mentioning before, with the latest ET version.
> >
> >
> > Thank you for your assistance!
> >
> >
> > Best,
> >
> >
> > Jordan
> >
> > ________________________________
> > From: Roland Haas <rhaas at mail.ubc.ca>
> > Sent: Friday, February 27, 2026 3:30:09 PM
> > To: Jordan Nicoules
> > Cc: Jordan Nicoules via Users
> > Subject: Re: [Users] Range error setting parameter, on cluster
> >
> >
> > CUIDADO: Email de um sistema externo. Cuidado com links, anexos e pedidos de dados/senhas.
> > CAUTION: Email from an external system. Be careful with links, attachments, and requests for data/passwords.
> >
> > Hello Jordan,
> >  
> > > Well, I don't know if the git hash corresponds to the ET release directly (I took it from the front page of the UsersGuide.pdf I have on my work station). The corresponding ET release should be ET_2024_05 I think.  
> >
> > The hash that is shown in the docs is a git commit hash of the Cactus
> > repository (the "flesh") so I can correlate with the release.
> >  
> > > I'm not sure this is (completely) a matter of version, since for a given version, I get different behaviors on different machines:
> > >
> > > - ET_2024_05: Deucalion --> range error ; work station --> fine
> > >
> > > - ET_2025_05: Deucalion --> range error ; work station, MesoPSL, MareNostrum --> all fine.
> > >  
> >
> > Ok, it working on a workstation and failing on another (or cluster,
> > https://en.wikipedia.org/wiki/Deucalion_(supercomputer) ?) is very strange.
> >
> > Would you be able to send me the option lists for those two and also
> > the files configs/<FOO>/config-data/make.config.defn ? The latter are
> > the fully parsed files that make constructs using the option list and
> > other information.
> >  
> > > Isn't ET_2025_05 the latest?  
> >
> > It is the latest yes.
> >
> > Yours,
> > Roland
> >
> > --
> > My email is as private as my paper mail. I therefore support encrypting
> > and signing email messages. Get my PGP key from http://pgp.mit.edu .  
> 
> 
> --
> My email is as private as my paper mail. I therefore support encrypting
> and signing email messages. Get my PGP key from http://pgp.mit.edu .

-- 
My email is as private as my paper mail. I therefore support encrypting
and signing email messages. Get my PGP key from http://pgp.mit.edu .