英国 BCS Fortran 年会講演
毎年講演には興味深いものがありますが、今年は歴史回顧的なもの、先人の労苦のねぎらい、ARM の Fortran 取り組み報告などのほか、題目だけですが Clive Page が I/O について講演したようです。最も興味深いのは J.K.Reid による Fortran 202yで導入予定の Generic についての講演で、日本の提案を紹介し米国案と比較しています。 なお ARM の資料では y は多分 8 とのこと。
なお J .K.Reid の米国案への見解は、スライド 15 頁に与えてあります。またスライド末尾に講演後の補足があります。
I confess to not fully understanding the proposal but it appears to me to be too complicated.
It is also far from complete.
I had hoped to show you how to create a generic procedure comparable to that available from the Japanese proposal, but I have failed to see how to do this.
Generic Programming (pdf) John Reid WG5 has accepted a proposal from the Japanese National Body for extending the generic capability that has been present in Fortran since Fortran 90. The programmer will be able to declare that a procedure is generic and has arguments that may have alternative types, kinds, or ranks. The compiler will generate those explicit versions that are invoked. It will solve a problem that my group has faced for more than 50 years. It began with having to write and maintain versions for both single and double precision and is worse now with more precisions of reals and several lengths of integers. This proposal will complement a much more ambitious US proposal that has been developed by a J3 subgroup since the 2019 WG5 meeting in Japan. This talk will explain the Japanese proposal and give a very brief summary of the aims of the US proposal.