3 ms·
I've been using it locally to write a few small scripts, nothing publicly available (yet). Traits and roles are really neat - they feel like a natural conclusi
by woodruffw 10y ago
I've been using it locally to write a few small scripts, nothing publicly available (yet).
Traits and roles are really neat - they feel like a natural conclusion to what Ruby started building upon with respect to metaprogramming natural syntax and encouraging developer happiness over strictness.
On the whole, however, many small parts of Perl 6 feel immature. There's no core JSON/YAML module, and there's no "official" module manager yet - just panda[1] and a few others.
Overall I'd recommend tinkering with it, but I don't think it's ready (as an ecosystem) for larger development. Hopefully soon, though!
[1]: https://github.com/tadzik/panda https://github.com/tadzik/panda
- liveoneggs 10y agothey've moved on to zef for modules. https://github.com/ugexe/zef https://github.com/ugexe/zef http://modules.zef.pm/ http://modules.zef.pm/
- gkya 10y agoBTW, the source for Panda is very clear and readable, I was just glancing it and am able to reason about it even though I only superficially know Perl 5 to write some dirty scripts and the code is sparsely commented.
- tadzik_ 10y agoThanks blushes :)
- mst 10y agoRakudo has JSON support built in, and Rakudo Star, the example batteries-included dist, has two additional JSON modules - and ships zef for module management, replacing panda. The ecosystem is still young, and "not ready" is a perfectly reasonable point of view, but you're using definitions of core/official that don't really apply - rakudo is the base compiler, not a complete release.
- woodruffw 10y agoI've been using my distro's Rakudo/Perl 6 package, which appears to be fairly out of date. That's probably why I couldn't find a JSON module or zef. Thanks for clarifying.
- b2gills 10y agoIt sounds to me like you would like a "batteries included" version of Perl 6. (include JSON/YAML in the runtime) No. We only include things as part of the language where the design of it is immediately obvious, and is useful for many applications. It also helps if it has been proven in other languages. So Promises(Futures) are in there, but many other things are not. You can do anything in a module, so write an external module for it instead. Otherwise we would have the problems where the best module for the job is not in the runtime, but everyone uses the one that is in the runtime. I have heard that Python has this problem with HTTP. Think of it like this, a module for doing X is included with the first version of a language. That means it is a module that was not written by an expert, because at that point in time there was no expert in the language. Everyone uses it though; as you don't have to install it, because it is already there. We don't have an official module manager, because we want all of them to be equally as official as every other one. It is a good thing too; as most people in the know use zef currently, rather than panda. If we had one that was official, it would have been panda. That's also why there is not, nor will ever be an official implementation. Any implementation that passes the test suite is just as official as every other one. (although there is only one currently) Basically you are asking for things to be included, that are specifically not included for very good reasons. The way that this is dealt with is by using the idea of distributions like "Rakudo Star" for Perl 6, or "Strawberry Perl" for Perl 5 on windows. That is an implementation of Perl 6 along with a set of useful modules.
- woodruffw 10y agoFirst of all, thanks for responding. I understand the hesitation of language maintainers with respect to adding complex modules like JSON and YAML, but I do think that there are some major benefits to doing so: * Users don't have to worry about inconsistent distributions. Following a test suite is one thing, but test suites are (almost) never comprehensive and will not catch programmer errors with respect to particular systems (e.g, LP64 and LLP64). Everything will work until it doesn't, and programmers will have to provide edge-case workarounds until individual distributions fix their implementations. The same goes for module managers. It's good to provide an open spec and encourage multiple frontends, but it's also good to provide a uniform interface for system administrators who just want to automate module installation for an application or framework. Python's easy_install wasn't ideal, but it provided exactly that sort of common denominator. "Batteries included" makes it sound like I want my language to come with a playground, which I think is an unfair way to characterize the expectation that Perl 6 provide feature parity with the languages it competes with (i.e., Ruby 2 and Python 3).