On Sun, 17 Dec 2017 23:40:43 +0100 Leon Timmermans <fawaka@gmail.com> wrote: > I think in this phase Paul's approach is the right one. Formalizing > such APIs is a good idea, but it having seem some real usage to proof > that they're the right APIs is valuable. If it all works out well > (which I suspect it will) then embedding the APIs in core is a > logical follow-up after a while. Indeed, as has been my plan. I've seen many cases both large and small (C99 and Scheme R6RS being two publicly-documented examples that come to mind) of APIs designed before solid use-cases have been laid down, and they've often ended badly. I want to avoid that here by first having a real use case that can be abstracted and generalised into an API design with at least one known usage. I have the intention here of implementing the async/await behaviour by *initially* whatever dirty tricks I have to make, to get it to work. Once it works, and from the outside appears to be neat and useful for CPAN modules, then the case can be examined to see if it feels sufficiently useful and universal to justify giving it greater support in core. I feel that my "suspend and resume a running PP sub" function pair may be the way to get that support into core, but I'm still sufficiently early in the experiment of the overall syntax and semantics to say for sure whether that will definitely be the case. I do understand that p5p@ may be concerned about future maintainability here - please appreciate that concerns me too. I don't want my users complaining that a subsequent version of perl breaks it, any more than you do :) I hope to be able to arrive at a situation that is mutually satisfactory to all. -- Paul "LeoNerd" Evans leonerd@leonerd.org.uk | https://metacpan.org/author/PEVANS http://www.leonerd.org.uk/ | https://www.tindie.com/stores/leonerd/Thread Previous | Thread Next