Re: An embedded DB engine

From: Richard Maine (nospam_at_see.signature)
Date: 01/02/04


Date: Fri, 02 Jan 2004 11:40:49 -0800

Ken Plotkin <kplotkin@nospam-cox.net> writes:

> My philosophy is to not arbitrarily impose my will and style on
> others. If I'm using a library written by someone else, I don't care
> what it's written in, as long as it's Fortran-callable.

Agree 100% with Ken here (and with the other things he said in
his post on this).

I've run into prejudices like this in the other direction - people
wasting a lot of time to rewrite libraries that I've done because
they don't want a solution in Fortran. They were making no
functional change at all - the library did what they wanted, and
even had a defined C API. They just had a prejudice against it
being in Fortran.

One of the stated reasons was that it would be easier to find
people to maintain a C version. This was pretty much belied when
they later came back and asked me to maintain their rewritten C
version (I refused because it takes me about an order of magnitude
to do things in C than in Fortran - whatever is true for other
people, I know that is true for me).

> as long as it's Fortran callable.

That's an important condition. If you want Fortran programmers
to use it, you have to define and support a Fortran API. You
can't just document a C API and expect Fortran programmers to
know how to call it - it typically won't even be possible without
writing at least some C for a wrapper, which wil be a significant
bar for some.

And supporting a Fortran API wil require documenting/testing
the particular compiler combinations that it works with, since
there are differences.

This is an area where I expect the F2003 interop to be of
substantial help - not yet, of course, because you can't assume
that all (or scarcely any) compiler support it. But someday.

-- 
Richard Maine
email: my last name at domain
domain: sumertriangle dot net


Relevant Pages

  • Re: Matlab Vectorisation Speed - How is it done in c++?
    ... Beating the performance of vectorized Matlab code is very ... Matlab makes calls to optimized C and Fortran libraries ... Use optimization level 3 on numerical code and level 2 on non- ...
    (comp.soft-sys.matlab)
  • Re: can somebody verify this C program which calls dsaupd_ ? (longish)
    ... >>>(terribly sorry for this long post which includes my two source files, ... > is NOT relevant to FORTRAN language newsgroup. ... More importantly, if your libraries compiled correctly, they were fine. ... > FORTRAN libs. ...
    (sci.math.num-analysis)
  • Re: Link problem with Compaq Visual Fortran
    ... > | yet one of the source codes is a Fortran program. ... > in Fortran and C run-time libraries (CVF uses both despite the fact ... GCG.lib: error LNK2001: unresolved external symbol ...
    (comp.lang.fortran)
  • NAG on Fortran for finance quants
    ... Malcolm Cohen of NAG has written an article on the advantages of a ... matrix-oriented language like Fortran 2003 for quantitative finance ... promoting Fortran in QF. ... adding NEW procedures ONLY to the F77 and C libraries. ...
    (comp.lang.fortran)
  • Re: C versus FORTRAN90
    ... FORTRAN90 version of these libraries. ... Are there other more subtle benefits for FORTRAN over C? ... If the answer to if f95 heavy, then maybe some sort of cost-benefit analysis of a wholesale switch from f95 to C at your company should be done to convince them otherwise. ...
    (comp.lang.fortran)