| wclodius@lanl.gov 2007-08-29, 10:07 pm |
| On Aug 29, 3:29 pm, Dan Nagle <danna...@verizon.net> wrote:
> Hello,
>
> The argument made below has been made before, and it's not quite
> right.
>
> The problem is that almost *nothing* in f08 depends upon anything
> in f03. Most f08 features could have been added to f95 directly.
> (Out of 40-some-odd features in f08, there's IIRC 2 depending
> upon something in f03.)
>
> How much experience do we need with f95 to add things?
>
> The real issue here isn't f08 at all. The real issue is that
> too much of what's in f03 doesn't have a strong cost-benefit ratio.
> That is, there's too much in f03 that while useful,
> doesn't have enough utility to justify it's addition to the language.
> A leading candidate here is DTIO, which has already generated
> many interpretation requests (which take time to resolve,
> because no one has the experience to do so!).
>
> What's left in f08 now is co-arrays and small stuff.
> How much experience with f03 do you need to add Bessel functions?
> Or to specify that the LGT() etc functions may take ASCII arguments
> (when the default characters are something else)?
To make this comparison more stark, how much of an effort would it
have been to add these to F77? Or F90 where the namespace control of
F90's modules helps in controlling the interactions of standard
libraries with the user's namespace. While the public domain libraries
are often cited as a strength of the language, the small size of the
standard libraries is a weakness. Their large libraries is a
significant reason for the popularity of commercial numerical
languages such as Matlab and IDL. It is too easy to concentrate on a
language's grammar, and neglect its vocbulary.
> <snip>
|