Around this time last year, Jonathan Worthington was writing their Advent Post called Reminiscence, refinement, revolution. Today, yours truly finds themselves writing a similar blog post after what can only be called a peculiar year in the world.
The Language
Visible Changes
The most visible highlights in the Raku Programming Language are basically:
last / next with a value
use v6.e.PREVIEW;
say (^5).grep: { $_ == 3 ?? last False !! True } # (0 1 2)
say (^5).grep: { $_ == 3 ?? last True !! True } # (0 1 2 3)Normally, last just stops an iteration, but now you can give it a value as well, which can be handy e.g. in a grep where you know the current value is the last, but you still want to include it.
use v6.e.PREVIEW;
say (^5).map: { next if $_ == 2; $_ } # (0 1 3 4)
say (^5).map: { next 42 if $_ == 2; $_ } # (0 1 42 3 4)Similarly with map, if you want to skip a value (which was already possible), you can now replace that value by another value.
Note that you need to activate the upcoming 6.e Raku language level to enable this feature, as there were some potential issues when activated in 6.d. But that’s just one example of future proofing the Raku Programming Language.
.pick(**)
The .pick(*) call will produce all possible values of the Iterable on which it is called in random order, and then stop. The .pick(**) will do the same, but then start again producing values in (another) random order until exhausted, ad infinitum.
.say for (^5).pick(* ); # 34021
.say for (^5).pick(**); # 340210214334210....Nothing essential, but it is sure nice to have 😀.
is implementation-detail trait
The is implementation-detail trait indicates that something that is publicly visible, still should be considered off-limits as it is a detail of the implementation of something (whether that is your own code, or the core). This will also mark something as invisible for standard introspection:
class A {
method foo() is implementation-detail { }
method bar() { }
}
.name.say for A.^methods; # barBUILDALLSubroutines and methods in the core that are considered to be an implementation-detail, have been marked as such. This should make it more clear which parts of the Rakudo implementation are game, and which parts are off-limits for developers (knowing that they can be changed without notice). Yet another way to make sure that any Raku programs will continue to work with future versions of the Raku Programming Language.
Invisible Changes
There were many smaller and bigger fixes and improvements “under the hood” of the Raku Programming Language. Some code refactoring that e.g. made Allomorph a proper class, without changing any functionality of allomorphs in general. Or speeding up by using smarter algorithms, or by refactoring so that common hot code paths became smaller than the inlinining limit, and thus become a lot faster.
But the BIG thing in the past year, was that the so-called “new-disp” work was merged. In short, you could compare this to ripping out a gasoline engine from a car (with all its optimizations for fuel efficiency of 100+ years of experience) and replacing this by an electrical engine, while its being driven running errands. And although the electrical engine is already much more efficient to begin with, it still can gain a lot from more optimizations.
For yours truly, the notion that it is better to remove certain optimizations written in C in the virtual machine engine, and replace them by code written in NQP, was the most eye-opening one. The reason for this is that all of the optimization work that MoarVM does at runtime, can only work on the parts it understands. And C code, is not what MoarVM understands, so it can not optimize that at runtime. Simple things such as assignment had been optimized in C code and basically had become an “island” of unoptimization. But no more!
The current state of this work, is that it for now is a step forward, but also a step back in some aspects (at least for now). Some workflows most definitely have benefitted from the work so far (especially if you dispatch on anything that has a where clause in it, or use NativeCall directly, or indirectly with e.g. Inline::Perl5). But the bare startup time of Rakudo has increased. Which has its effect on the speed with which modules are installed, or testing is being done.
The really good thing about this work, is that it will allow more people to work on optimizing Rakudo, as that optimizing work can now be done in NQP, rather than in C. The next year will most definitely see one or more blog posts and/or presentations about this, to lower the already lowered threshold even further.
In any case, kudos to Jonathan Worthington, Stefan Seifert, Daniel Green, Nicholas Clark and many, many others for pulling this off! The Virtual Machine of choice for the Raku Programming Language is now ready to be taken for many more spins!
The Ecosystem
Thanks to Cro, a set of libraries for building reactive distributed systems (lovingly crafted to take advantage of all Raku has to offer), a number of ecosystem related services have come into development and production.
zef ecosystem
The new zef ecosystem has become of age and is now supported by various developer apps, such as App::Mi6, which basically reduces the distribution upload / commit process to a single mi6 release↵. Recommended by yours truly, especially if you are about to develop a Raku module from scratch. There are a number of advantages to using the zef ecosystem.
direct availability
Whenever you upload a new distribution to the zef ecosystem, it becomes (almost) immediately available for download by users. This is particularly handy for CI situations, if you are first updating one or more dependencies of a distribution: the zef CLI wouldn’t know about your upload upto an hour later on the older ecosystem backends.
better ecosystem security
Distributions from the older ecosystem backends could be removed by the author without the ecosystem noticing it (p6c), or not immediately noticing it (CPAN). Distributions, once uploaded to the zef ecosystem, can not be removed.
more dogfooding
The zef ecosystem is completely written in the Raku Programming Language itself. And you could argue that’s one more place where Raku is in production. Kudos to Nick Logan and Tony O’Dell for making this all happen!
raku.land
raku.land is a place where one can browse the Raku ecosystem. A website entirely developed with the Raku Programming Language, it should be seen as the successor of the modules.raku.org website, which is not based on Raku itself. Although some of the features are still missing, it is an excellent piece of work by James Raspass and very much under continuous development.
Not forgetting the past
“Those who cannot remember the past are condemned to repeat it.” George Santanaya has said. And that is certainly true in the context of the Raku Programming Language with its now 20+ year history.
Permanent Distributions
Even though distributions can not be removed from the zef ecosystem, there’s of course still a chance that it may become unavailable temporarily, or more permanently. And there are still many distributions in the old ecosystems that can still disappear for whatever reason. Which is why the Raku Ecosystem Archive has been created: this provides a place where (ideally) all distributions ever to be available in the Raku ecosystem, are archived. In Perl terms: a BackPAN if you will. Before long, this repository will be able to serve as another backend for zef, in case a distribution one needs, is no longer available.
Permanent Blog Posts
A lot of blog post have been written in the 20+ year history of what is now the Raku Programming Language. They provide sometimes invaluable insights into the development of all aspects of the Raku Programming Language. Sadly, some of these blog posts have been lost in the mists of time. To prevent more memory loss, the CCR – The Raku Collect, Conserve and Remaster Project was started. I’m pretty sure a Cro-driven website will soon emerge that will make these saved blog posts more generally available. In the meantime, if you know of any old blog posts not yet collected, please make an issue for it.
Permanent IRC Logs
Ever since 2005, IRC has been the focal point of discussions between developers and users of the Raku Programming Language. In order to preserve all these discussions, a repository was started to store all of these logs, up to the present. Updating of the repository is not yet completey automated, but if you want to search something in the logs, or just want to keep up-to-date without using an IRC client, you can check out the experimental IRC Logs server (completely written in the Raku Programming Language).
Looking forward
So what will the coming year(s) bring? That is a good question.
The Raku Programming Language is an all volunteer Open Source project without any big financial backing. As such, it is dependent on the time that people put into it voluntarily. That doesn’t mean that plans cannot be made. But sometimes, sometimes even regularly, $work and Real Life take precedence and change the planning. And things take longer than expected.
If you want to make things happen in the Raku Programming Language, you can start by reporting problems in a manner that other people can easily reproduce. Or if it is a documentation problem, create a Pull Request with the way you think it should be. In other words: put some effort into it yourself. It will be recognized and appreciated by other people in the Raku Community.
Now that we’ve established that, let’s have a look at some of the developments now that we ensured the Raku Programming Language is a good base to build more things upon.
new-disp based improvements
The tools that “new-disp” has made available, haven’t really been used all that much yet: the emphasis was on making things work again (after the engine had been ripped out)! So one can expect quite a few performance improvements to come to fruition now that it all works. Which in turn will make some language changes possible that were previously deemed too hard, or affecting the general performance of Raku too much.
RakuAST
Jonathan Worthington‘s focus has been mostly on the “new-disp” work, but the work on RakuAST will be picked up again as well. This should give the Raku Programming Language a very visible boost, by adding full blown macro and after that proper slang support. While making all applications that depend to an extent on generating Raku code and then executing it, much easier to make and maintain (e.g. Cro routing and templates, printf functionality that doesn’t depend on running a grammar every time it is called).
More Cro driven websites
It looks like most, if not all Raku related websites, will be running on Cro in the coming year. With a few new ones as well (no, yours truly will not reveal more at this moment).
A new language level
After the work on RakuAST has become stable enough, a new language level (tentatively still called “6.e”) will become the default. The intent is to come with language levels more frequently than before (the last language level increase was in 2018), targeting a yearly language level increase.
More community
The new #raku-beginner channel has become pretty active. It’s good to see a lot of new names on that channel, also thanks to a Discord bridge (kudos to Wenzel P.P. Peppmeyer for that).
The coming year will see some Raku-only events. First, there is the Raku DevRoom at FOSDEM (5-6 February), which will be an online-only event (you can still sign up for a presentation or a lightning talk!). And if all goes ok, there will be an in-person/online hybrid Raku Conference in Riga (August 11-13 2022).
And of course there are other events where Raku users are welcome: the German Perl/Raku Workshop (30 March/1 April in Leipzig), and the Perl and Raku Conference in Houston (21-25 June).
And who knows, once Covid restrictions have been lifted, how many other Raku related events will be organized!
Finally
This year saw the loss of a lot of life. Within the Raku Community, we sadly had to say goodbye to Robert Lemmen and David H. Adler. Belated kudos to them for their contributions to what is now the Raku Programming Language, and Open Source in general. You are missed!
Which brings yours truly to instill in all of you again: please be careful, stay healthy and keep up the good work!
Raku Advent Calendar 
One thought on “Day 25 – Future-proofing the Raku Programming Language”