Fortran - Slides for presentation on F2003

Author

Slides for presentation on F2003

Bart Vandewoestyne

2005-09-07, 6:59 pm

Dear all,

A little while ago, I've mentioned my idea on giving a seminar at
my department on the new features of Fortran 2003 because I think
a lot of people still don't know what it's all about, what the
possibilities of Fortran 90/95 are and what the possibilities of
Fortran 2003 will be.

I asked my PhD promotor and it shouldn't be a problem giving a
seminar on that somewhere in the (far?) future. I'm in no hurry
to create this presentation, so I would like to do it thoroughly,
work on it at some moments in my free time, during w

ends
etc... and I would like to create a presentation that I can make
publicly available so it can be used and adapted by others to
spread the word about Fortran 2003.

Therefore, dear group, i need your help. I want to do this step
by step, doing several iterations, basically in the following
form:

Step 1: i put an initial idea online about the content of the
presentation

Step 2: the group gives feedback and gives me comments,
suggestions, ...

Step 3: i change the presentation according to the group's
recommendations, add content, and try to make it something
decent.

Step 4: the group gives me feedback again

Step 5: i change again, thereby refining the presentation, making
it better and better :-)

Step 6: etc... etc... etc...

At the end, we should have a very clear and structured
presentation about the new features of Fortran 2003 and although
I would have done most of the typing, i absolutely wouldn't mind
if others used the same presentation with some possible minor
changes.

So... I've created a basic structure for now... the PDF and .tex can
be found at

http://www.cs.kuleuven.ac.be/~bartv/f2003/seminar2.pdf
http://www.cs.kuleuven.ac.be/~bartv/f2003/seminar2.tex

I am using the LaTeX Beamer Class to create the presentation:

http://latex-beamer.sourceforge.net/

So, as to you, the audience, i would ask:

* Please also read the remarks at the top of the .tex file on
intended audience and duration.

* What do you think of the current structure? Are there things
to add, things to remove? Let's not go too much into detail
for now, only give your opinion on the basic structure of the
talk... the details are for later.

* If you download and change the .tex file yourself, then you can
always send me a patch... this saves me some typing ;-) I
cannot however guarantee that I will use your patch as this
will depend on my and on comp.lang.fortran's opinion.

Hoping to get a lot of good feedback,
Bart

--
"Share what you know. Learn what you don't."

[JvO]

2005-09-07, 6:59 pm

Bart Vandewoestyne schreef:

> Dear all,
>
> A little while ago, I've mentioned my idea on giving a seminar at
> my department on the new features of Fortran 2003 because I think
> a lot of people still don't know what it's all about, what the
> possibilities of Fortran 90/95 are and what the possibilities of
> Fortran 2003 will be.
>

About sheet # 8:

ANSI X3.9-1978 was also ISO 1539-1980

Sheet # 9:

I happen to posess a copy of one of the drafts called Fortran 8X.
Very interesting to read ...
Fortran 8X was indeed not a standard but a preliminary name for Fortran
90. Hoping to get a series 66 77 88 ?

[JvO] aka Jan van Oosterwijk

> Hoping to get a lot of good feedback,
> Bart
>
> --
> "Share what you know. Learn what you don't."

beliavsky@aol.com

2005-09-07, 6:59 pm

Bart Vandewoestyne wrote:
> Dear all,
>
> A little while ago, I've mentioned my idea on giving a seminar at
> my department on the new features of Fortran 2003 because I think
> a lot of people still don't know what it's all about, what the
> possibilities of Fortran 90/95 are and what the possibilities of
> Fortran 2003 will be.
>
> I asked my PhD promotor and it shouldn't be a problem giving a
> seminar on that somewhere in the (far?) future. I'm in no hurry
> to create this presentation, so I would like to do it thoroughly,
> work on it at some moments in my free time, during w

ends
> etc... and I would like to create a presentation that I can make
> publicly available so it can be used and adapted by others to
> spread the word about Fortran 2003.

This project sounds interesting. One place to work on and "publish" a
free textbook/tutorial is the Wikibooks at
http://en.wikibooks.org/wiki/Main_Page . The September 2005 book of the
month was on Ada 2005. I wrote 0.1% of a Fortran textbook at
http://en.wikibooks.org/wiki/Programming:Fortran , and you are welcome
to take over (and don't need anyone's permission, anyway).

Regarding your conclusion that

"Fortran is great for numerics and number-crunching stuff, but
a lot less well-suited for interactive point-and-click GUI's,
networking, . . .! use the right tool for the right job!",

this may a property of the relative lack of available libraries than of
the language itself.

Richard E Maine

2005-09-07, 6:59 pm

In article <1126120640.011701.269420@o13g2000cwo.googlegroups.com>,
"[JvO]" <jvo_36@hotmail.com> wrote:

> Fortran 8X was indeed not a standard but a preliminary name for Fortran
> 90. Hoping to get a series 66 77 88 ?

The "standard" after-the-fact joke on this was that the X in 8X was a
Roman numeral.

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain | experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

Bart Vandewoestyne

2005-09-07, 6:59 pm

On 2005-09-07, beliavsky@aol.com <beliavsky@aol.com> wrote:
>
> Regarding your conclusion that
>
> "Fortran is great for numerics and number-crunching stuff, but
> a lot less well-suited for interactive point-and-click GUI's,
> networking, . . .! use the right tool for the right job!",
>
> this may a property of the relative lack of available libraries than of
> the language itself.

True. I did not exactly know how to formulate this... what i
meant is that if you only stick to standard-conforming Fortran to
keep your software portable, then you cannot use vendor-specific
libraries and extensions that allow you to create more graphical
user intefaces or do networking stuff etc...

I just might want to add 'Standard' in front of that item, or if
anybody knows how to formulate this more clearly, feel free to do
so...

Regards,
Bart

--
"Share what you know. Learn what you don't."

James Giles

2005-09-07, 6:59 pm

[JvO] wrote:
> Bart Vandewoestyne schreef:
.....
> I happen to posess a copy of one of the drafts called Fortran 8X.
> Very interesting to read ...
> Fortran 8X was indeed not a standard but a preliminary name for
> Fortran
> 90. Hoping to get a series 66 77 88 ?

Fortran 8X was very much a different language than F90. When
the public review for 8X was negative, there was a massive
reworking of most of the new features. In retrospect, many
(perhaps most) of the 8X features were actually better than
we ended up with. In any case, F8X was neither a separate
standard nor a preliminary name for F90, it was a significantly
different language that was discarded.

--
J. Giles

"I conclude that there are two ways of constructing a software
design: One way is to make it so simple that there are obviously
no deficiencies and the other way is to make it so complicated
that there are no obvious deficiencies." -- C. A. R. Hoare

Richard E Maine

2005-09-07, 6:59 pm

In article
<f_HTe.202098$5N3.183196@bgtnsc05-news.ops.worldnet.att.net>,
"James Giles" <jamesgiles@worldnet.att.net> wrote:

> In any case, F8X was neither a separate
> standard nor a preliminary name for F90, it was a significantly
> different language that was discarded.

Well... the various F8X documents were, in fact, drafts of what became
f90. I'd agree that some of the changes were pretty significant. Whether
they were so large as to constitute a different language is a judgment
call that I don't think I'll argue with you. But it is a matter of fact,
not of judgment, that the f90 standard was produced by a process of
making changes to the drafts informally referred to as F8X. It was not
done by discarding the old document and starting over.

I've got copies of some of the F8X drafts, and it isn't hard to find
matching sections in them and the F90 standard.

I'd say that the number of changes was relatively small in some sense.
I've seen a summary that was only about a handful of bullet items. But
some of them were in areas that had ramifications all over the document
(for example, KINDs replacing the older concept).

See Metcalf's inside story of the tale. (I don't have the URL handy, but
it has been posted here before and is probably findable with google).

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain | experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

James Giles

2005-09-07, 6:59 pm

Richard E Maine wrote:
....
> I've got copies of some of the F8X drafts, and it isn't hard to find
> matching sections in them and the F90 standard.

It isn't hard to find things that changed *significantly* either
(except for the F77 portions). Many of those changes were
not improvements. (I'm understating my opinion here.)

--
J. Giles

"I conclude that there are two ways of constructing a software
design: One way is to make it so simple that there are obviously
no deficiencies and the other way is to make it so complicated
that there are no obvious deficiencies." -- C. A. R. Hoare

Richard Maine

2005-09-07, 9:56 pm

In article <1126124280.123862@seven.kulnet.kuleuven.ac.be>,
Bart Vandewoestyne <MyFirstName.MyLastName@telenet.be> wrote:

> what i
> meant is that if you only stick to standard-conforming Fortran to
> keep your software portable, then you cannot use vendor-specific
> libraries and extensions that allow you to create more graphical
> user intefaces or do networking stuff

And of course that is also true of C, C++, and most other languages.
Standard C doesn't have any GUI libraries either. So I don't think that
catches quite the core point.

Brooks Moses

2005-09-07, 9:56 pm

Bart Vandewoestyne wrote:
> On 2005-09-07, beliavsky@aol.com <beliavsky@aol.com> wrote:
>
> True. I did not exactly know how to formulate this... what i
> meant is that if you only stick to standard-conforming Fortran to
> keep your software portable, then you cannot use vendor-specific
> libraries and extensions that allow you to create more graphical
> user intefaces or do networking stuff etc...

I think the fundamental point here is that Fortran is designed to be a
_very_ portable language -- the standard makes remarkably few
assumptions about the hardware and input/output systems available.

GUI programming, on the other hand, is quite hardware-specific, and in
most cases quite OS-specific as well.

These are inherently contradictory concepts. One can get around the OS
restrictions to a fair extent by using libraries that abstract the
interactions with the OS, but one is still making signficant assumptions
about the underlying hardware capabilities that are not present in the
Fortran standard.

So, the corrolary to this fundamental point is that, in order to do GUI
programming, one must use libraries that are less portable -- and, in
some cases, much less portable -- than the underlying language.

As Richard says, this is true of most other languages as well. I think
the additional point is that most of the libraries are written for
interactions with C and/or C++, and thus using them with Fortran tends
to be a bit more difficult....

- Brooks

--
The "bmoses-nospam" address is valid; no unmunging needed.

glen herrmannsfeldt

2005-09-08, 3:57 am

Richard Maine wrote:

> In article <1126124280.123862@seven.kulnet.kuleuven.ac.be>,
> Bart Vandewoestyne <MyFirstName.MyLastName@telenet.be> wrote:

[color=darkred]
> And of course that is also true of C, C++, and most other languages.
> Standard C doesn't have any GUI libraries either. So I don't think that
> catches quite the core point.

Well, except that many GUI libraries are written in C or C++, just
like many numerical libraries are in Fortran. To call one from the
other is a little more complicated, not counting the F2003 features.

-- glen

Richard Maine

2005-09-08, 3:57 am

In article <1ZadnTgNJ7XJMILeRVn-sA@comcast.com>,
glen herrmannsfeldt <gah@ugcs.caltech.edu> wrote:

> Richard Maine wrote:
>
>
>
>
> Well, except that many GUI libraries are written in C or C++, just
> like many numerical libraries are in Fortran. To call one from the
> other is a little more complicated, not counting the F2003 features.

Indeed. That's more the point to be made. I just don't think that Bart's
words said that. That is quite likely to be what he meant, I think, but
we were trying to help get the wording right. I don't think we were
really arguing about what the situation is, but rather how to state it
well. You've provided a plausibly good statement of it, I'd say.

Pierre Asselin

2005-09-08, 6:59 pm

Bart Vandewoestyne <MyFirstName.MyLastName@telenet.be> wrote:

> [ ... ]
> I asked my PhD promotor and it shouldn't be a problem giving a
> seminar on [F2003] somewhere in the (far?) future.
> [ ... ] Therefore, dear group, i need your help.

From your .tex file,

-----
% TARGET AUDIENCE
% ---------------
% People in the area of computer science that are comfortable with software
% development, programming, numerical computing,... but with little or no
% knowledge about Fortran or people having only knowledge up to the Fortran 95
% standard.
-----

In my humble opinion...

For that kind of audience, I would put the C interoperability
features at the very top of the list. They probably all know C,
even the ones that hate it, and they all know how to glue C to
their preferred programming language. Knowing that they can do
the same with Fortran would be a big deal to them.

Next on the list would be the type-bound procedures and other
similar features that support (limited ?) object-oriented programming.
Also procedure pointers.

If they don't know much about Fortran 95 you should also talk about
modules and pointers. In particular, how they are *not* analogous
to similar-sounding constructs in other languages ! This should
be number 2 on the list, right after C interop.

--
pa at panix dot com

Brooks Moses

2005-09-09, 3:57 am

Pierre Asselin wrote:
> For that kind of audience, I would put the C interoperability
> features at the very top of the list. They probably all know C,
> even the ones that hate it, and they all know how to glue C to
> their preferred programming language. Knowing that they can do
> the same with Fortran would be a big deal to them.
>
> Next on the list would be the type-bound procedures and other
> similar features that support (limited ?) object-oriented programming.
> Also procedure pointers.
>
> If they don't know much about Fortran 95 you should also talk about
> modules and pointers. In particular, how they are *not* analogous
> to similar-sounding constructs in other languages ! This should
> be number 2 on the list, right after C interop.

Those sound useful. (I haven't looked at the actual slides.)

The thing with those points, though -- they're useful as far as they go,
but they're mostly talking about how Fortran can be "as good as"
something else. It's also important to talk about how it's better.

The key points on how Fortran is _better_ than other languages, from my
perspective, are twofold: First, the array handling -- whole-array
syntax, sane passing of multidimensional arrays as function arguments,
elemental functions, and so forth. Second, the "allocatable"
functionality, and the fact that it can do a substantial amount of what
pointers are often used for, but it simply doesn't leak memory.

- Brooks

--
The "bmoses-nospam" address is valid; no unmunging needed.

Bart Vandewoestyne

2005-09-09, 7:56 am

On 2005-09-07, Bart Vandewoestyne <MyFirstName.MyLastName@telenet.be> wrote:
>
> A little while ago, I've mentioned my idea on giving a seminar at
> my department on the new features of Fortran 2003 because I think
> a lot of people still don't know what it's all about, what the
> possibilities of Fortran 90/95 are and what the possibilities of
> Fortran 2003 will be.
> [...]

I've added some things, trying to listen to what the group
already mentioned in this thread. The update is at

http://www.cs.kuleuven.ac.be/~bartv/f2003/seminar2.pdf

The LaTeX source together with the images I'm using in the
presentation is at

http://www.cs.kuleuven.ac.be/~bartv/f2003/seminar2.zip

To be continued... in the meanwhile, early feedback is of course
welcome. It's better to change it now in this early stage than to
change it once it's all written down :-)

Regards,
Bart

--
"Share what you know. Learn what you don't."

Michael Metcalf

2005-09-10, 6:58 pm

"Richard E Maine" <nospam@see.signature> wrote in message
news:nospam-667B93.14060807092005@news.supernews.com...
>
> See Metcalf's inside story of the tale. (I don't have the URL handy, but
> it has been posted here before and is probably findable with google).
>
> --

It's not a URL, but the Foreword of "Numerical Recipes in Fortran 90". Note
too that we published two editions of "Fortran 8x Explained", and the
relevant Tables of Contents can be used to follow the officially published
drafts and to compare them to the final standard as described in "Fortran 90
Explained".

Regards,

Mike Metcalf

P.S. I've been silent here for about two months as I've spent that time
moving from Berlin to New York. The installation of internet in our flat was
delayed at the last moment when my young son tipped orange juice over the
laptop's keyboard, bur we're up and running as of yesterday.

Brooks Moses

2005-09-10, 6:58 pm

Michael Metcalf wrote:
> P.S. I've been silent here for about two months as I've spent that time
> moving from Berlin to New York. The installation of internet in our flat was
> delayed at the last moment when my young son tipped orange juice over the
> laptop's keyboard, bur we're up and running as of yesterday.

Welcome back! (And may your future computer setup be less exciting! :)

- Brooks

--
The "bmoses-nospam" address is valid; no unmunging needed.

Sponsored Links