Re^8: Perl 6: Managing breakages across Rakudo versions

[Anonymous Monk]

in reply to Re^7: Perl 6: Managing breakages across Rakudo versions
in thread Perl 6: Managing breakages across Rakudo Star versions

I will give you a frank feedback on what I think should be the way forward for managing things in Rakudo

1. No more major rewrites, as they break a lot of code and a lot of catching up is required in the months that follow.

2. Implementation for at least one backend complete, which is backwards compatible, has basic libraries and usable documentation.

3. Path towards making Perl 5 and CPAN compatible defined, and planned to achieve over a period of time.

[cavac]

I agree.

I speak only for myself here as someone who gets paid to write Perl code.

Just give me a Perl6 implementation i can depend on. I need to start writing scripts and modules and ditribute them on CPAN. And i need some reasonable guarantees that i don't have to do major rewrites every few months just to keep my code working.

Rakudo doesn't have to be "feature complete" or be "highly optimized". To attract developers (who may write Perl code for a living), it needs to be stable enough that it's worth the (company) time to start writing code for it that it more complex than "Hello World" and more useful than self-test scripts.

Stop breaking things because you "need to optimize". Frankly, unless Perl6 gets at least one stable, extensible Interpreter (e.g. not major rewrites every time you feel like it), there won't be a big community writing actual, useful code for it that starts to attract even more developers. Ergo it wont run any code that needs an optimized interpreter...

Really, give me something i can show to my boss as a reasonable good choice to implement software with that will run for years, maybe a decade.

[jdporter]

I agree! Except that I really am in no rush. I want to see a production release as soon as possible, but I don't want to see these pre-production releases dressed in the trappings of production releases (e.g. "stability policies" and other service level agreements) because that actually harms the reputation of Perl. I agree with Anonymous Monk: until such time as Perl 6 is really production ready, break anything and everything necessary to get it there. The more you break now, the less you'll break later. Presumably.

[raiph]

Being free to break things without regard for users both helps and hinders Perl 6 and Rakudo in getting to the next level.

It helps in the sense it lets the designers and implementors move more freely and quickly. The Rakudo compiler is developed in this spirit.

Breaking things hinders Perl 6 and Rakudo to the degree there's a need for many more folk who successfully deploy Perl 6 in a production setting before considering calling Perl 6 "production ready" and folk who might do so are averse to signs of excessive instability. The Rakudo Star bundle is developed for these folk.

To encourage folk to try Perl 6 in production settings, and respond to complaints about rewrites, Patrick considers it necessary to at least have explicit public understanding about managing breakages across versions of Rakudo Star.

(Fwiw I would say the Parrot team made the mistake of prematurely deploying an onerous stability policy but the Perl 6 and Rakudo teams have not.)