First official release of LLVM Flang

[deng]

Congratulations to the team, this has been a long time coming. I still think that modern Fortran is actually a great language to write numerical code, especially when doing lots of linear algebra. Granted, it was many years ago, but I still remember struggling with C++ and libraries like Eigen, and one day, confronted yet again with agonizing slow compile times and error messages that look like binary, I ditched C++ for good and moved to Fortran95. Not only could I pretty much copy&paste a lot of stuff from my Matlab prototype, the resulting binary was actually faster than C++ with Eigen.

Not sure if I would use it today for new projects, probably Julia would be the better choice nowadays.

[accurrent]

IF we can get modern fortran working on top of GPUs via spir-v that would be really awesome. Modern Fortran is so nice for numerics.

[pandemic_region]

> Whilst many alternative programming languages have come and gone, it [Fortran] has regained its popularity for writing high performance codes.

I don't understand why sometimes people pluralize "code". It sounds a bit silly but maybe it's just me.

[enriquto]

it's a fairly common usage in numeric computing. If you read, for example, the wikipedia entries for "computational fluid dynamics" you'll see that they consistently speak of "codes" when referring to programs.

https://en.wikipedia.org/wiki/Computational_fluid_dynamics

[thenoblesunfish]

It's because the types of things people write in Fortran (high performance science codes, for example) tend to be monolithic, single-purpose programs. It comes from a time when a code really was basically one compilation unit (and doing that is such a nice simplification that I support it, for science). With code written for the web, shared through package managers, etc. it makes more sense to use the uncountable noun instead of the countable one.

[bradrn]

As someone doing research in physics, I’ve noticed this usage before. It seems fairly frequent outside CS [EDIT: and as the sibling comment says, specifically in numerical computing]. From what I’ve gathered, for them ‘code’ has become a count noun, such that ‘a code’ means something like ‘a piece of code’ or even ‘a program’, and the plural ‘codes’ follows from that.

[gpvos]

I'm seeing it here for the first time, and from multiple people. Maybe it's specific to the Fortran community?

[jmclnx]

Congratulations, next is COBOL ? I am serious, we really need a free COBOL compiler.

Yes, GNU now has a front end for COBOL, so LLVM turn. Maybe IBM and the Navy Department will help.

[lproven]

Good news!

https://www.theregister.com/2025/03/13/gcc_15_is_close/

[aragilar]

There doesn't appear to be a link to the release notes, it would be nice to know what are the current limitations.

[pklausler]

Language-wise, it's all of F'2018 less coarrays and "LEN" derived type parameters, plus a pile of portable &/or popular language extensions.

[wiz21c]

How compatible is it with the current code bases developped under proprietary compilers ?

[pklausler]

My approach throughout was to maximize portability of existing code to this new compiler. The list of extensions that are supported is quite long (https://flang.llvm.org/docs/Extensions.html), and the general policy is to support anything that people need so long as the feature is well defined and portable among compilers that support it.

[melodyogonna]

It says in the article that some of these companies contribute their test suite

[HexDecOctBin]

Now someone go convince the C committee to stop trying to turn C into Fortran.

[froh]

Q from out of the loop: what are you referring to? and if you could enlighten me how that then is a problem?