Thank you very much! And thank you for the link! This will make some good bed time reading 
1 Like
In ResearchGate, I have just noticed a huge surge in the reads of “The State of Fortran” last week. Have you any idea of what could explain such a peak? 
4 Likes
Well I think that’s the numFOCUS Organization. I shamelessly included the paper as one of the contributions of Fortran-lang’s members. The date matches perfectly too.
4 Likes
I also linked to it in this post. 
https://degenerateconic.com/a-modern-fortran-scientific-programming-ecosystem.html
8 Likes
It’s @jacobwilliams’s post that made it to the front page of Hacker News. Typically it gets a few thousand reads, of which perhaps 10% clicks on the link to the article, so around 300 reads is about right.
5 Likes
LLVM flang Open source, Apache-2.0 Under development, full support for parsing F2018
What does “full support for parsing F2018” mean? By some definition, /usr/bin/true also has full parsing support for F2018.
In case of LLVM flang the compiler is afterwards emitting valid Fortran source showing that the round trip from source to internal representation to source works. While one could argue that is a feature /usr/bin/cat has as well, let’s focus on the development effort behind these projects instead.
My point is that this is not a useful statement from a user perspective. I do a huge amount of serious application programming in Fortran and do not care one iota about whether the parser doesn’t crash while doing nothing.
1 Like
I’m looking at the development of LLVM based Fortran compilers from a user perspective as well. So far none of the available or in development ones can compile any of my projects. Still I’m excited for the first one to reach production level stability, having a robust parser is just one milestone for such an effort.
1 Like
The statement “full support for parsing F2018” is not directed towards the end user. It’s a development milestone for compiler authors and people that have been keeping up to date with the progress of LLVM flang. As @awvwgk mentioned LLVM flang is not a stable compiler for end users. Nonetheless, when stable it will be a great alternative to gfortran and lfortran (also in active development).
4 Likes
The statement “full support for parsing F2018” is not directed towards the end user.
Who do you think the audience of this paper is?
People, Organisations and Institutions not using Fortran have read the paper and reached out, so definitely not just end users.
4 Likes
To whom is the statement “full support for parsing F2018” directed, i.e. who is going to find that statement relevant and informative?
I assert that you have made it far too easy for readers to misjudge the readiness of LLVM Fortran for actual use, in order to make a statement that is only interesting to people close to the project, who do not need to be told this.
1 Like
I don’t completely follow the argument here and why is there so much weight on this part of the classification? The full statement reads Under development, full support for parsing F2018, which in my opinion does not make any wrong promises to end users. Since the paper describes the current state of the language, the development of compilers, even if those are currently in an early stage, is something I consider an important information on a language.
7 Likes
@JeffH is your point that we oversold LLVM Flang? We used this phrase from some materials published by Flang developers, I can’t remember if it was the J3 committee paper progress, or some other material. If we did, then I apologize. Our intention was to represent it as best as we could.
Can you give us a better phrase to use instead? (For other papers or communication, going forward.)
Disclaimer: I develop LFortran, a competing compiler to Flang. So I am in a conflict of interest here. I am personally trying to be as objective as I can, both on this discourse, and at the fortran-lang.org page (I created the initial compiler section) and discussed with many interested parties to send us how they want to describe the compiler. You are in a conflict of interest too, working for a company that develops Flang. But ultimately we are all on the same team: team Fortran. So that is our goal here. So if you can help us represent Flang so that it is fair, that would be awesome.
9 Likes
Yes, I worry that saying LLVM Fortran parses all of Fortran 2018 is going to cause users to try it and be disappointed. Many Fortran users do not understand the nuance of parsing versus actually generating working binaries.
The HPC community is full of examples of projects that got too much attention too soon and suffered as a result. There are people who still refused to use NWChem a decade after the first public release, because the first time they tried it, there were bugs.
I don’t have a suggestion to improve this that fits into a table, unfortunately.
2 Likes
I have just found that August IEEE news about The State of Fortran. It is a very imperfect resume, as for example it tells Fortran lacks a central website and an online community, or a community compiler, but does not say it has been or is being fixed. But that news does exist and has a good title…
The authors of the paper can try to reach out to the IEEE Computing Society Team and provide an overview of where things are and request an edit of the blogpost that adopts a more positive light reflecting the changes and ongoing developments in the Fortran “ecosystem”, as opposed to the misleading take with “lacks” this and that.
4 Likes
the Fortran package manager the team is developing provides a build system for Fortran and simplifies the compiling process.
Who is “the team” here? I haven’t looked at the complete paper here, but the front page there reads like a crummy AI lifting of content from a handful of recent papers on the topic.
1 Like
Yes, either a journalist reading quickly the paper or an AI… (but I thought AI were better at resuming). Anyway, it is rather disappointing for an IEEE site.
Note that you can fill the form to download the paper, but you will have to check the box “I am not a robot”