It's clear that James Giles thinks there's Something Wrong with the way
Fortran standards are developed, and therefore Something Wrong with the
standards themselves.
At least some of the process is dictated by ISO, ANSI and INCITS. WG5 and
J3 therefore have no control over at least those aspects of the process.
During the twenty years it took to convince my employer to support my
participation in J3 and WG5 meetings, I held opinions similar to Giles's
concerning How the Committees Work and How the Committees Ought to Work.
Upon attending a few meetings, I found that those opinions were largely
the result of my profound ignorance.
I remain slightly disappointed that J3 and WG5 cannot do more by way of
correspondence between meetings, but I now see that the magnitude of
inter-meeting progress I hoped for before actually participating is not
really possible.
Giles is disappointed about the Private Nature of J3 and WG5 deliberations.
As many others have pointed out, this is simply not the case. Since he
is clearly mistaken on this point, perhaps the source of his discontent is
that he isn't personally apprised of every decision that takes place during
the development of a standard. As anybody who has participated in any
kind of organization knows, one of the most difficult things is to recruit
a secretary to record the minutes. The J3 minutes record the results of
votes on directions papers should take between revisions, and on whether
the resulting papers are finally approved. Discussions are not scribed.
J3 would welcome Giles to come to meetings and scribe discussions for
posting on this mailing list or in news groups.
Giles is disappointed about the level of publicity concerning possible
directions revision of the standard might take. When my account on the
computer that was in the J3 mailing list was cancelled without my knowledge
some time around 1994, I didn't blame J3 for not finding me. I lamented
that the adminstrator for that computer hadn't told me that mail was being
forwarded from that account. But in the end I realized it was my own damn
fault for not keeping myself informed. Giles may wish that Somebody on J3
would keep him and this mailing list and some news group informed, but
consider that most members of J3 are sent there by their employers. For
most of those delegates, it's all they can do to secure support to attend
the meetings -- and some delegates can't even get support to attend every
meeting. To my knowledge, nobody who attends has a title at his workplace
something like "Full-time J3 representative and J3 outreach officer." Most
of the delegates do their inter-meeting homework at home, not at the office.
J3 would welcome Giles to come to meetings, and then publicize the proceedings
in this and other fora, or at least to read the meeting papers and minutes,
and summarize their results for this and other fora. Perhaps the reason
Giles doesn't do it is that his employer doesn't provide support to do it.
Yes, there are some delegates who attend J3 meetings at their own expense.
We admire them for doing so, and appreciate their contribution. Even they,
however, have demands upon their time beyond J3 work, such as operating their
own businesses. To my knowledge, none are supported by Charitable Foundation
Grants.
Giles appears to be disappointed at the direction that the 2003 revision
took, offering criticism that new features were untried. It would be
interesting to see an itemized list of the new features, as described in
ftp://ftp.nag.co.uk/sc22wg5/N1551-N1600/N1579.ps.gz, annotated with which
of them is not either (a) an obvious extension of an existing Fortran
feature, (b) an obvious method to solve obvious problems, or (c) something
that's already been well-tried in other languages, tailored to fit in the
existing Fortran framework. The only one that comes immediately to mind
as not falling into any of those categories is user-defined derived-type
input/output.
An example of (a) is parameterized derived types, and the supporting
infrastructure for it, including extensions to the ALLOCATE statement.
An example of (b) is the IOMSG= specifier in an I/O statement. An
example of (c) is the object-oriented programming facility, which is
consciously based on how it is done in Simula and Beta. Sure, there were
numerous ways that each new feature could have been implemented, but the
final form was the result of discussion and debate. A glimpse of the
discussion and debate appears in the meeting papers, but the really intense
work was done face-to-face at the J3 and WG5 meetings. It was difficult
for me to imagine how important the in-person interaction is until I
actually participated.
--
Van Snyder | What fraction of Americans believe
[log in to unmask] | Wrestling is real and NASA is fake?
Any alleged opinions are my own and have not been approved or disapproved
by JPL, CalTech, NASA, Sean O'Keefe, George Bush, the Pope, or anybody else.
|