Skip to content

FORTRAN Newsgroup

Home arrow Computational arrow FORTRAN arrow FORTRAN Newsgroup


FORTRAN Newsgroup
Discussion about FORTRAN.

  • Re: Help with nntp server (OT)
    On Wed, 20 Aug 2008 09:49:12 -0700 (PDT), michaelmetc...@compuserve.com
    posted:
    news.individual.net was the answer for me. It costs something like fifteen
    bucks for a year but is worth every penny. It takes the spam and porn
    right out of usenet. The folks at the freie universitaet in Berlin run it,

  • wholesale all kind of shoes
    SELL:NIKE AIR FORCE 1 [link]
    AIR FORCE 1 LIGHT UP
    AIR DUNK SB
    NIKE DUNKS
    AF1 25TH
    AIR FORCE 1 25TH
    AIR Jordan Fusion 1
    AIR Jordan Fusion 2
    AIR Jordan Fusion 3
    AIR Jordan Fusion 4
    AIR Jordan Fusion 5
    AIR Jordan Fusion 6
    AIR Jordan Fusion 7
    AIR Jordan Fusion 8
    AIR Jordan Fusion 9
    AIR Jordan Fusion 10

  • Re: compiler flag question
    I was thinking of having access via whatever preprocessor macro to
    the active compiler options. As OpenMP does require a specific
    macro to be present, so much the better :).
    (If OP's compiler can't handle preprocessor macros, a solution
    would be to include a library with dummy routines)
    Regards,
    Arjen

  • Re: compiler flag question
    OpenMP standard requires the _OPENMP macro, so it seems cpp style
    conditional compilation is at least a de facto requirement for OpenMP.
    Conditional inclusion of omp functions is one of the most common uses of
    _OPENMP. For example, I want to see in my report whether OpenMP is
    active, with how many threads.

  • Re: Surprise
    The main language used for all sorts of applications, as well as many
    utilities used by the OS, or as a suite of tools and library routines
    provided by the OS for use by applications. That's what *we* used the
    term for, at any rate (USAF weather computing facility).
    The applications written in Fortran included number-crunching forecast

  • Re: compiler flag question
    Ah, well preprocessor-defined macros ought to do what you want,
    but you have to define them together with the OpenMP compiler option,
    as there is no way to get that automatically. At least not standard.
    (I got the impression that you would be solving different things
    or using different algorithms.)
    Though, perhaps the OpenMP routine omp_get_max_threads() could

  • Re: compiler flag question
    I basically want to have different output to certain files depending
    on how the user compiles the code.
    I just gave the loop problem as an example. I'm also incorporating
    some other features that I want to enable by compiler flags.

  • Re: compiler flag question
    How about
    imax = 5
    !$ imax = 10
    do i=1, imax
    :
    end do
    Regards
    tomgu...@hotmail.com schrieb:

  • Re: compiler flag question
    Out of curiosity: why is the loop different if you use OpenMP
    than if you don't - especially as this is a compile-time
    decision, not part of the input for the problem.
    Regards,
    Arjen

  • Re: compiler flag question
    You can use COCO if you want a Fortran-only solution
    Regards,
    Arjen

  • Re: compiler flag question
    If machine dependent, use
    call sub_openmp ( parms )
    call sub_other (parms )

  • Re: common data in fortran77
    EQUIVALENCE can avoid passing massive number of separated variables by
    using array. However, an additional include file is necessary which
    defines macros for all the separated variables. In this way, one will
    not worry about situations like removing /adding a variable from/to
    the array, which will change the array index for the separated

  • Re: common data in fortran77
    Yes. Exactly.
    What I need is
    (a) a header file that declares all variables
    (b) a subroutine that sync my variables to the latest status
    Then, in all my functional subroutines, I can include the header file
    and call the sync subroutine.
    It works, :-)
    Thank you all.
    Cheng'an

  • Re: ordering integer array
    In article ,
    The gfortran developers don't expect an ordinary user to read the
    source code. Heck, at least one former gfortran developer has come
    to realize that ordinary user typically don't read the manual, too. :)
    Actually, gfortran's random_number() will use up to 3 independent

  • Re: zipcode selection sort in MR&C
    [snip]
    Input data records are of the form:
    Program text:
    The output file contains 99 ios because program execution falls
    through to statement 99.
    How about something like this?
    program zippy
    use sort
    implicit none
    integer, parameter :: array_size = 100
    type (address), dimension (array_size) :: data_array

  • Re: compiler flag question
    On Wed, 20 Aug 2008 12:42:03 -0800, glen herrmannsfeldt posted:
    Yikes. You can do that with C, but would you?

  • Re: ordering integer array
    On Wed, 20 Aug 2008 04:52:18 +0000 (UTC), Steven G. Kargl posted:
    You must play different trivia games than do I. I've never hit the science
    and math category and been asked "how many random_seed calls do you have to
    make for an effective fortran pseudorandom implementation?"
    I see in here some of the same information as was in gfortran.pdf, to wit:

  • Re: zipcode selection sort in MR&C

    T. Boone may have a residence in Houston, I don't know, but he lives in
    the TX panhandle.
    Indianapolis may be humid, but Houston would probably win that contest.


  • Re: zipcode selection sort in MR&C
    On Tue, 19 Aug 2008 21:03:22 -0800, glen herrmannsfeldt posted:
    Thanks, Glen. I had two errors going at the same time. One was that I
    misspelled addresses.txt when I named the actual file (Germans use one d in
    address). That was why windows was creating another one. The other was
    that the addresses needed to formatted differently, as you indicated.

  • Re: Compaq Visual Fortran
    dpb wrote:
    There were early rumors that HP might be considering continuing active
    development of CVF for a period. I don't know whether that was real or
    whether it would have violated some contractual obligations.

  • Re: Surprise
    General purpose means in part that it can serve successfully in multiple
    problem domain areas. I merely provided an example beyond "number
    crunching.
    Anyway, uncle.

  • Re: Compaq Visual Fortran
    ...
    ...
    That part I didn't know...I had presumed "the compiler" staying implied
    the source as well. I guess that pretty effectively limited the HP
    chances for support/development/upgrades or did they get to fork the source?

  • Re: Surprise
    Initially, you were talking about "general purpose programming", not
    "systems programming".
    I guess this comes down to whether a program written in Fortran using
    non-standard extensions can support the statement that Fortran is a
    good general purpose programming language (GPPL). So basically a
    matter of definition of general purpose programming.

  • Re: ordering integer array
    On Tue, 19 Aug 2008 11:10:11 +0200, Kurt Kallblad posted:
    I modified your source slightly and tried it on silverfrost and gfortran.
    What follows is the source, the output for gfortran and then the output for
    silverfrost.
    program test
    real :: x(6); integer :: i
    integer, allocatable :: old1(:), old2(:)

  • Re: Help with nntp server (OT)
    In article , Gary Scott
    1&1, one of the more popular providers in Germany, provides the
    news.online.de NNTP server, which works fine. I've used it for years.
    They don't advertise it much, but it is there and problems are fixed if
    one complains (for example, a moderated group was marked as unmarked),

  • Re: Surprise
    I meant that I was the one who exaggerated. :)
    The word "intended" was the choice of Mr. Luka Djigas. I just tried to
    soften his statement, keeping the wording and to emphasize that yes,
    Fortran can be used for different tasks, even systems programming, but
    that would most likely, I would even say, inevitably involve using non-

  • Re: Surprise
    You seem to be equating ability to interoperate with any random other
    programming language as an essential component of "system programming".
    It isn't. If the OS API is written in the same language or compatible
    with, then ANY compatible language can be used for "system programming".
    And there are many examples of operating systems in which Fortran was

  • Re: Surprise
    I don't think the use of the "mis" prefix is right. This is my
    understanding based on experience and substantiated by what you said.
    This is how I read it.
    Victor.

  • Re: Surprise
    I'd say that's quite a misunderstanding.

  • Re: Surprise
    "Intended" is a questionable word to use in such contexts, but if one is
    being precise, so far as I can tell, it is humans that have intents
    (well, except for things like intent(in) :-). If you want to use the
    word "intended" (and it was your choice of word), then I don't see how
    you can find a more definitive statement of the intent of one of the

  • Re: Surprise
    Well, that statement might be overly exaggerated in terms of what
    Fortran is "intended" for, but what I see is that Fortran is primarily
    used for doing heavy computations. Sure enough, my experience is not
    as extended as yours, yet I see what I see.
    Victor.

  • Re: Defined operators and underscore
    I don't recall the underscore reason either.
    The digit restriction isn't "really" an ambiguity. Given something
    like
    123.e4.xyz blah blah blah
    it's "ambiguous" whether .xyx is the start of an operator that has
    123.e4 as it's first operand or whether e4 is the operator and 123 is
    the left operand. However, for a syntacticly correct blah blah blah

  • Re: Surprise
    So pretty much, to use Fortran for "general purpose programming",
    whatever that may be other than number crunching, one would either
    have to use extensions or call libraries written in other languages.
    So I'd rather say that you have "general purpose" compilers that allow
    you to perform the above mentioned tasks rather than a general purpose

  • Re: Help with nntp server (OT)
    Tastes good too.

  • Re: Compaq Visual Fortran
    Terence, I find that statement quite offensive as well as wildly inaccurate.
    Richard Maine explained it well and I will not elaborate.
    I am glad to help people with CVF issues if it seems important and I can do so
    easily. In this thread, the side question about installing upgrades did not
    rise above that threshold, and I was quite busy all last week, representing

  • Re: Help with nntp server (OT)
    I'm not sure how effective it is outside the U.S., but I've found
    Motzarella to work very well for me.

  • Re: Compaq Visual Fortran
    Correct, though the source code for CVF also moved to Intel.

  • Re: Surprise
    Since I participated in development of several of the Fortran standards
    (and was even editor of 2), and I personally know that I intended it for
    other things as well, that's about as direct a disproof of that
    statement as there can be.

  • Re: common data in fortran77
    No, the F77 standard is quite clear - if a local variable, initialised
    by a DATA statement is assigned a different value, it becomed undefind
    on exit from the route. (17.3.6.b)
    Dave Flower

  • Re: Surprise
    I'm using it to access I/O hardware through port read/writes, data
    acquisition, low level serial bus (MIL-STD-1553) programming, and
    similar "systems programming" tasks. I find that it serves these roles
    every bit as well as C, with only a single language extension
    (integer/cray/whatever pointers) and some compiler directives. F2003

  • Re: Help with nntp server (OT)
    ....
    Well, this is no criticism of Maine - I'm sure he is reporting
    exactly what he sees. But, I didn't have to figure out anything.
    I found an article I wanted to respond to and pressed "reply".
    At that point an editable window appeared with the article
    to which I was responding quoted within it (properly shifted

  • Re: compiler flag question
    You can't do that in Fortran, or pretty much any
    compiled language. (I have seen it done in BASIC).
    But you say compiler flag. If you use the C preprocessor,
    you can do
    do i=1,10
    do i=1,5
    where OPENMP is a preprocessor symbol, defined either
    with #define, or a compiler option such is -DOPENMP

  • Re: common data in fortran77
    I am not so sure what you are trying to do, but...
    If you have a large number of separate variables that you
    need to pass around to many routines, you could put them in
    a structure, though that requires features past Fortran 77.
    For Fortran 77, you can EQUIVALENCE them to array elements,
    and pass the array around. They must be the same type, though.

  • Re: Surprise
    Meaning the OS APIs were written in Fortran (and assembly) and tailored
    specifically for Fortran. Some OS' even carried over Fortrans concept
    of logical file numbers and similar features in the JCL.

  • Re: Surprise
    I disagree. Although, your definition of "general purpose language" is
    different.
    Victor.

  • Re: Surprise
    I guess a more correct statement would be "standard Fortran was never
    intended for anything but number crunching." Extensions provided the
    functionality that made it possible to use Fortran for other tasks.
    IMHO.
    Victor.

  • Re: Surprise
    On Wed, 20 Aug 2008 10:53:48 -0500, Gary Scott
    Please, elaborate. What do you mean under "main systesm programming
    language" ?
    Best regards
    Luka

  • Re: Fortran Forum Recommendation
    I guess another point...some of the code snippets are in an atrocious
    style with no indention, no line spacing, etc. Recommend a standardized
    clean, easier to read style for code snippets.

  • Re: J3 Responses to Public Comments
    Yes, I think that's what I said. Some could be modeled the old way.
    Others simply cannot. All can be modelled the new way.
    1. Consistency is good as a general principle. Perhaps not as a hobbling
    mandate (cite Emerson), but certainly as a guide. Consistency is
    possible and indeed quite easy in this case, though only if one sticks

  • Re: Help with nntp server (OT)
    On Wed, 20 Aug 2008 10:05:54 -0700, Richard Maine
    wrote in <1ilyuas.bzxp8m15cmyqwN%nos... @see.signature>:
    Many people over here also use individual, and swear by it
    (although their preferred payment method has drawn criticism lately).
    Luckily I have my academic connections (although Brunel figures that since