(j3.2006) (SC22WG5.5914) Preview of possible feature survey

Bill Long longb
Wed Jul 12 14:52:25 EDT 2017


I?m still concerned about how the class of ?users? is identified.  Asking on lists like the Fortran-90 email list, or (especially) comp.lang.fortran would not represent a useful cross-section of ?users?.   Asking for input from some forums could result in a load of junk that will consume a lot of committee time to read and respond to.   It would be more useful to look at the core users/authors of major applications (user sites like NCAR, NERSC, Archer, LRZ, ECMWF, the Met Office, ..., or applications like GAMESS, NWChem, ...) who might be influenced to stick with Fortran based on what changes we make.  Performance of the generated code is always high on the list, but ease of code maintenance, or clarity of expressing efficient algorithms that apply to a large class of problems also seem important.   Or maybe we should look at large scientific codes written in other languages (mainly C++, it seems) and ask the authors what Fortran lacks that motivated using a different language.  

It might also be helpful to limit the number of proposals from any one individual, which would force him/her to focus on what is really important.  A number like 5 seems reasonable. I don?t recall more that that number of acceptable proposals from anyone in the past.  It could help cut down on the amount of commitee time spent weed-whacking.

Cheers,
Bill


> On Jul 12, 2017, at 9:51 AM, Steve Lionel <steve at stevelionel.com> wrote:
> 
> John,
> 
> Thanks for your comments.
> 
> Simply asking for suggestions doesn't provide us with information on what users think is important. We ARE considering (or have considered in the past) many of these items, and my intent is to populate the list with new suggestions as they come in. The survey allows people to suggest additional items.
> 
> We are not at all bound by this, but having ranked-choice voting encourages users to recognize that there are tradeoffs and that we can't do everything.
> 
> Steve
> 
> On Wed, Jul 12, 2017 at 5:19 AM, John Reid <John.Reid at stfc.ac.uk> wrote:
> Steve,
> 
> Sorry to be slow in commenting on this. I am not happy about have a list
> like this. It suggests that the committee is seriously considering all
> these and that some sort of vote will determine which are chosen. I
> think it would be better just to have the open invitation to make a
> suggestion, with a full explanation of how they would use it and why it
> would make a big difference in their application.
> 
> Cheers,
> 
> John.
> 
> Steve Lionel wrote:
> > At Garching we agreed to survey the user community to see what features
> > they wanted added to Fortran. I volunteered to create such a survey.
> >
> > I have put together a sample survey which you can preview at
> > https://www.surveymonkey.com/r/Preview/?sm=mQpu1omN0yqfQzTHlOh1EINqgBiSYtQgDwCuXU3_2BipuWire738v3KF9WXKQ5T0Q2
> > You can comment in the preview (at least I think you can) by clicking on
> > the little speech box icons, or you can send me your comments.
> >
> > In this iteration I have asked the user to rank the features, which is
> > helpful in establishing priorities, but an alternative is star ratings
> > or asking to rank each option on a 0-5 scale. These would get a
> > different perspective, but I am uncertain which is more useful to us.
> >
> > I look forward to your comments. Also if you have suggestions to add to
> > the list of features, let me know. Everything about this is up in the air.
> >
> > On a semi-unrelated note, on a suggestion from John and after
> > consultation with Malcolm, I have registered the domain wg5-fortran.org
> > and am setting up a web site under that domain. This would have the WG5
> > content currently hosted by NAG. When that is done, we can have the NAG
> > URL redirect to the new one for a while.
> >
> > Steve
> >
> >
> > _______________________________________________
> > ukfortran mailing list
> > https://lists.accu.org/mailman/listinfo/ukfortran
> _______________________________________________
> J3 mailing list
> J3 at mailman.j3-fortran.org
> http://mailman.j3-fortran.org/mailman/listinfo/j3
> 
> 
> 
> -- 
> .
> _______________________________________________
> J3 mailing list
> J3 at mailman.j3-fortran.org
> http://mailman.j3-fortran.org/mailman/listinfo/j3

Bill Long                                                                       longb at cray.com
Principal Engineer, Fortran Technical Support &   voice:  651-605-9024
Bioinformatics Software Development                      fax:  651-605-9143
Cray Inc./ 2131 Lindau Lane/  Suite 1000/  Bloomington, MN  55425





More information about the J3 mailing list