Evolution of Fortran standards

The page counts of Fortran standards (parenthesized below) are a crude measure of language complexity. One can study the evolution of Fortran by reading documents at J3 and WG5.

Fortran 2018 (646)
Fortran 2008 (623)
Fortran 2003 (583)
Fortran 95 (376)
Fortran 90 (315)
Fortran 77 (185)
Fortran 66 (39)

7 Likes

The Fortran 202x draft (2021-12-27 version) is 681 pages.

2 Likes

I was approached by the ISO for exactly the same thing. I no longer provide any links.

2 Likes

Does Fortran need to be liberated from ISO? Is it just an impediment at this point? Weird archaic processes, people having to sneak around to get “draft” versions of documents written 100% by volunteers, very strange and confusing two-level committee structure that seems to not be able to work in any sort of modern way, etc., etc. Would more people be willing to volunteer to work on Fortran standards if all of that was swept away?

5 Likes

That would certainly be a tragedy, but if true it would ultimately create a double standard of how ISO treats Fortran and other languages like C/C++.

I will admit I have little knowledge on the matter, but I was intrigued as to why the ISO standards cost so much and why they are labelled as open-access even though you have to pay for access.

In trying to understand the following I did some reading and I realised that C/C++ does the same thing. They publish their standards with ISO but they also freely release their drafts in open-std.org. I am in no position to know if ISO has contacted people to take down links to standard drafts but given that there are numerous places online like Stack Overflow, cppreference.com and GitHub where links to all the C++ standard drafts are being displayed I would say that it is quite unreasonable for ISO to request J3 to take down theirs.

3 Likes

I’m wondering about the exact same thing. When considering different programming languages for a product, ISO-certification is nowhere near being a requirement for my work. I appreciate that others may have different requirements though, so I’m genuinely curious to hear argument in advantage of ISO.

The way I see it there is a a grave being dug for the traditional design-by-committee languages (C, C++) by the newer and more open languages like Rust and Golang. As it stands today I see hardly any arguments in favour of them unless you have a large legacy codebase there. To be brutally honest I’m not sure Fortran is much different. I maintain a large codebase already in Fortran so turning down Fortran is a choice I’m rarely is able to make anyways.

Given the current amount of compiler bugs in any fortran compiler I’ve used - open source or not - I believe free access to the standards are absolutely crucial not only for a single compiler, but for the survival of the language as a whole.

3 Likes

Another option would be to contact a lawyer to clarify what is actually permitted.
ISO has an interest in selling the official version, so no wonder that they ask individuals to take down links to the draft versions. Whether this request is backed up by law is not said.
For academic papers, it is obvious that publishers have the copyright to the final printed version, but they can’t prohibit uploads of preprints to arxiv etc. Some countries even require by law that research funded by public money needs to be openly available.

1 Like

We have actually looked at the feasibility of divorcing from ISO, and perhaps organizing under IEEE. The problem, as I recall, is that some countries national bodies have a tight relationship with ISO and that such a separation would be impractical.

@kargl is correct that ISO is very protective of its copyrights and has been known to hunt down what it considers copyright violators. It’s why J3 is very careful to not claim to separately offer copies of the standard, but rather the “committee interpretation reference document”, which is subtly different from the ISO standard (it has line numbers, for one!)

Please don’t kick the bear.

10 Likes

That is very interesting, “committee interpretation reference document” sounds like some very expensive legal advice. Do you know by any chance if the drafts offered by C++'s WG21 are any different?

1 Like

I have changed the title thread and removed links to specific standards, and I will not tweet such links.

2 Likes

That isn’t something I have looked at.

For Fortran, J3 posts drafts that it produces, but any documents submitted to ISO are not published in places we (J3/WG5) control. You’ll note that on the WG5 web site, you can find all WG5 documents except drafts submitted to ISO and final standards. The process we have works for us and hasn’t yet subjected us to complaints from ISO - I’d like to keep it that way.

3 Likes

Care to explain a bit more? I don’t understand how any country’s national bodies are related to the design of a programming language.

1 Like

Here, “National Bodies” refers to country groups devoted to creating and maintaining standards; it’s an ISO term. J3 (INCITS PL22.3) is the US National Body for Fortran and is tasked with the technical development of the standard. Some of the other national bodies participating in Fortran development are DIN (Germany), JIS (Japan) and BSI (UK).

For J3 in particular, it’s INCITS, the parent organization, that has close ties with ISO.

1 Like

Thanks!

1 Like

I think C++ has a lot more people (in general; and thus also regarding those) who were chafing against those constraints, but fundamentally what’s possible for C++ in an ISO context should be possible for Fortran (and others) as well?

The C++ draft is even hosted on Github these days, and there’s also a fully navigable (though non-official) website.

When J3 tried to separate from INCITS (WG5 did not try to separate from ISO), IEEE wouldn’t touch us without getting permission from INCITS, which was not forthcoming. If IEEE had become the parent body, the international standard would most likely still have been published by ISO.