Loading...
You are not logged-in Login/Register





  • Posts   Search Threads
  • brianlammJune 5, 2007 9:36 AM PDT   
    When are the REAL Fortran 2003 Capabilities to be Provided?

    It's been almost four years since the f2003 standard was officially adopted, yet Intel still lags far behind all other major Fortran compiler vendors in providing the new functionalities.  NAG 5.1 is superior in that regard.

    The f2003 implemented by IVF is, so far, miniscule at best.

    Things like the full capability of the ALLOCATE statement (new = typed and sourced allocation, IVF has neither); parameterized derived types; type extension; CLASS; abstract interfaces; procedure pointers; type-bound procedures; deferred attribute; nopass and pass attributes; IEEE 754 arithmetic exceptions, etc; the list of useful and powerful f2003 capabilities not implemented by IVF goes on and on.

    And Intel provides incorrect information for a f2003 feature Intel says is implemented: derived type components may be allocatable.  That's incorrect and Intel should make it clear only derived type components that are arrays may be declared allocatable, so that's yet another very useful f2003 thing Intel has not implemented: allocatable scalars, which is huge when one doesn't need the added functionality (and needless headaches) of using pointers.

    So, when is Intel going to get earnest about implementing useful f2003 capabilities?

    Brian Lamm



    Ronald W. Green (Intel)June 6, 2007 4:56 PM PDT
    Rate
     
    		 #1
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?

    Brian,

    I will try to contact you offline so we can discuss Fortran 2003 features.  After our discussion I will post a reply for the rest of the group.

     

    ron




    How to attach files to a posting
    Private posts and threads how-to


    brianlammJune 7, 2007 12:17 PM PDT
    Rate
     
    		 #2 Reply To #1
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?

    Ron,

    Sounds good to me.  You should have my registered email address on file.

    Thanks,

    Brian



    brianlammJune 12, 2007 7:43 AM PDT
    Rate
     
    		 #3 Reply To #2
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?

    I would just like to add Ron called me and after the conversation was over I realized, once again, why I stick with Intel: the people behind and on the "scene".

    Brian Lamm



    jeanvezinaJune 21, 2007 8:26 AM PDT
    Rate
     
    		 #4 Reply To #3
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?

    Any news about the Fortran 2003 implementation roadmap ?

    When a fully compliant compiler will be available ? 2008 ? 2009 ? 2010 ?

    The order of priority should be: Object Oriented features first, user-defined derived type

    I/O last. The COM wizard and Module Wizard should be changed to use F2003 syntax.

    Best regards,

    Jean Vezina

     

    P.S. The rather ironic thing about this is that the Fortran 2008 standard may well become official before complete Fortran 2003 compilers are even released.



    Steve Lionel (Intel)June 21, 2007 8:42 AM PDT
    Rate
     
    		 #5 Reply To #4
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?
    We can't offer a roadmap or timeframe.  We are committed to fully supporting F2003, but that's all we can say.  And yes, it is indeed likely that F2008 will be finalized before there are "mainstream" F2003 compilers available.  That point is not lost on some on the standards committee.

    I'd appreciate more detail on your suggestion about the COM and Module wizards. Care to give an example (in general terms)?


    Steve

    Attaching or including files in a post
    Doctor Fortran blog
    @DoctorFortran on Twitter

    jeanvezinaJune 22, 2007 6:22 AM PDT
    Rate
     
    		 #6 Reply To #5
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?

    Dear Steve,

    Currently the Module Wizard generates a wrapper around COM,OLE, or Active X objects. Calling the generated routines is not as straighforward and convenient as it is in other languages such as Visual Basic or C#.

    With Fortran 2003, a notation closer to the normal "object notation" could in principle be used.

    For example: setting a property:  object%property = ... instead of CALL AUTOSETPROPERTY.

    Calling a method:  call object%method(...) instead of calling a generated subroutine with awkward arguments.

    Object finalization (destructors) is also now part of Fortran 2003. Currently, all the housekeeping has to be done by the programmer to avoid memory leaks. In Visual Basic, C++, and C#, this is done much more automatically.

    Again, that depends on the long term view of Intel about Fortran. Does Intel sees it as a general application development language (Digital and after Compaq did some good steps in that direction) or just a high performance computational tool where all the GUI and O/S interactions are handled by other languages, leaving the Fortran part just for calculations ?

    Best regards,

    Jean Vezina

     

     



    jimdempseyatthecoveJune 22, 2007 7:31 AM PDT
    Rate
     
    		 #7 Reply To #5
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?

    >> And yes, it is indeed likely that F2008 will be finalized before there are "mainstream" F2003 compilers available.  That point is not lost on some on the standards committee.

    Ahh, there's the problem... The standards committee is moving too fast ;)

    Jim



    Blog: The Parallel Void
    www.quickthreadprogramming.com

    Steve Lionel (Intel)June 22, 2007 7:36 AM PDT
    Rate
     
    		 #8 Reply To #6
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?
    Jean,

    Thanks for taking the time to give us your thoughts on this. I can't make any promises, but we'll certainly consider something along these lines. 


    Steve

    Attaching or including files in a post
    Doctor Fortran blog
    @DoctorFortran on Twitter

    djpinksterFebruary 18, 2008 1:24 PM PST
    Rate
     
    		 #9 Reply To #8
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?
    Hi, it's been a while since the last post on this thread, and I'm wondering what the progress is on this topic in VF10.1. Looking at the "new features" list I see something on Fortran 2003 features but nothing on classes, inheritance or methods (/procedures) as part of types. Can't find anything in the IVF (10.0) help-files either. Has any of this been implemented, or is this on a release-schudule sometime this year?

    From my point of view, if these features aren't in the language it's really difficult to lift my collection of Fortran (77/90) sources to the next level of (re)useability and maintainability so I'm really excited to hear what the plans are for IVF in this respect.

    Regards, Dirk-Jan


    abhimodakFebruary 18, 2008 2:01 PM PST
    Rate
     
    		 #10 Reply To #9
    Re: When are the REAL Fortran 2003 Capabilities to be Provided?
    I am hoping that Ron will inform us about "the minutes" of his discussion.....

    Abhi


Forum jump:  

Intel Software Network Forums Statistics

17,025 users have contributed to 48,321 threads and 172,753 posts to date.

In the past 24 hours, we have 16 new thread(s) 57 new posts(s), and 54 new user(s).

In the past 3 days, the most popular thread for everyone has been How to manage rounding by IVF ?? The most posts were made to Most likely, the issue is that The post with the most views is Optimalization of sine function\'s taylor expansion

Please welcome our newest member redfruit83


For more complete information about compiler optimizations, see our Optimization Notice.