|Think about Loose Coupling|
A wholly inadequate reply to an Anonymous Monkby pmichaud (Scribe)
|on Apr 22, 2010 at 21:36 UTC||Need Help??|
An Anonymous Monk on perlmonks.org recently started a discussion on "The current state of Perl 6" in which he (she? they?) offers his views on various aspects of Perl 6 development. Personally, I find nearly all of Anonymous Monk's comments in the thread to be so devoid of logic and so far removed from reality/facts that it's hard to imagine providing an adequate reply to them.
Beyond that, I have other priorities in my life right now. Uppermost is that my family is going through a very difficult period at present -- I need to take care of them first. Rakudo and Perl 6 come in at a somewhat distant second to that. Responding to garbage being spouted by anonymous person(s) probably ought to not even be anywhere on my radar.
But at least one post in the thread bugs me sufficiently that I'd like to respond, even if it's an inadequate response. And I don't want my response buried in a thread somewhere, so it gets its own post.
Anonymous Monk writes:
Oh I'll tell you how you do that [write a grammar engine capable of supporting Perl 6]. It's very simple. You get people skilled for this exact task ! Those skills are acquired in universities(preferably good ones) where professors teach courses like "Formal Languages and Automata" or "Compiler theory". If you have a bunch of people who are open-source contributors but don't have the knowledge or haven't studied the right things ... [t]hey don't know it in theory and they're trying to put it in practice! (emphasis in original)
Who does Anonymous Monk think is working on Perl 6? Personally, I have a Ph.D. in Computer Science and taught university language courses for nearly 14 years. Damian Conway certainly also has a fairly strong university background (Monash University) and understands the theory and practice of language and compiler design. Jonathan Worthington has a degree from the University of Cambridge, where he specialized in compiler construction and formal languages. Larry Wall has a degree in Natural and Artificial Languages with two years of graduate work in Linguistics at U.C. Berkeley. Many of our other key contributors also have degrees and backgrounds in language theory and practice. I'd be willing to compare the academic credentials of this group against any other dynamic language team Anonymous Monk might care to postulate.
If you can't get real compiler people to use Perl6 and help with it, the average open-source rookie won't be able to deal with this.
Somehow I have trouble classifying Larry, Damian, Allison, etc. as being "not real compiler people". And they along with many other Perl 6 contributors have a lot of experience in nurturing open source developers and projects. If you feel Larry et al. are indeed unqualified to be working in the field of programming language design and implementation, you probably don't want to be using Perl at all, much less Perl 6.
People like Anonymous Monk spew lots of opinions about how long they think Perl 6 development should take, and then speculate on the reasons why it has taken longer than they estimate it should be taking. The speculations that crop up most often are things like "the people involved are clueless about language design/implementation" (see above), "the design process itself is flawed", "they're building on an inadequate toolset like Parrot", etc. For such people it's easy to toss out random thoughts about "the problems with Perl 6" without bothering to look at the obvious evidence to the contrary, as Anonymous Monk does above. Indeed, I'm often amused when people suggest that we should be doing things that we're already doing.
Returning to the topic of developing a grammar engine, which Anonymous Monk claims above as "very simple" and just needing "people skilled for this exact task", it's interesting to contrast his opinions with the actual development of the Perl 6 standard grammar (STD.pm6). I think STD.pm6 is also indicative of the challenges confronting all Perl 6 implementors. Consider:
I think one could consider this to be an almost perfect situation for developing a new language implementation -- an experienced language designer/implementor working without significant external restrictions on top of an advanced programming platform like Perl 5. Yet it's been three years since Larry started working on this implementation of the standard grammar and parser, and while STD.pm6 is very powerful, it's also certainly not "finished", and has been through a number of significant refactors in its development. This says to me that the amount of time and effort involved is more likely due to the sheer audacity and ambition of what Perl 6 seeks to accomplish.
(Some will undoubtedly look at the above and knee-jerk respond with something like "If it's taken three years just to create the parser, then finishing a compiler will take far longer still and therefore Perl 6 will never see the light of day." This argument ignores the fact that other pieces are being implemented in parallel with Larry's work, that software development is not a strictly sequential process, and the obvious fact that there are already "working" Perl 6 implementations available.)
Anyone who thinks that Perl 6 is fundamentally based on traditional compiler construction techniques taught in universities frankly has no clue as to what a fundamental paradigm shift Perl 6 represents to language design and implementation. It's this fundamental change that ultimately gives Perl 6 its power, but it's also why Perl 6 development is not a trivial exercise that can be done by a few dedicated undergraduates. As TimToady on #perl6 says, "We already have lots of those kinds of languages."
Personally, I'm impressed and honored to be associated with the people who are working on Rakudo and Perl 6. I understand that people are frustrated (and even feel burned by) the long wait for something as cool as Perl 6; I share the frustration and like to think that I'm doing something constructive about it. But I also find the vast majority of suggestions, comments, and conclusions coming from Perl 6's anonymous detractors to be (1) things we've already done or are doing, (2) ungrounded in reality, (3) in direct contradiction to reasonably observable facts, or (4) attempts to discredit a product they have little interest in ever using themselves. And far too many of the comments, like the ones I've highlighted in this post, are so easily refuted with just a little bit of fact digging and common sense, it's often hard to believe anyone can be seriously making them in the first place. Yet there they are.
Returning to my original theme, I think my response here is inadequate because it leaves so many other of Anonymous Monk's claims in the thread unrefuted. I could undoubtedly spend many more hours analyzing and responding to the many fallacies and untruths in the thread, but frankly I don't believe that's the best use of my time. People such as Moritz Lenz, chromatic, Michael Schwern, and others are also writing extremely well-reasoned posts refuting the garbage, for which I'm grateful, but it's far easier for Anonymous Monk and his like to spin garbage than it is for a small number of us to clean up after it. And it does need to be cleaned up, otherwise it festers and results in even more public garbage that someone has to clean up.
I hope that this post will at least encourage more people in the Perl community to critically examine the things they hear and read regarding Perl 6, especially when coming from sources with no apparent standing or reputation within the community. And maybe a few more people will even assist in publicly refuting the garbage ("many hands make light work"), so that the sloppy thinking, analysis, and dialogue that people like Anonymous Monk post doesn't spread to infect all of Perl.
P.S.: Some may reasonably conclude from what I've written above that Perl 6 is somehow "aiming too high", that our goals should be scaled back to make something ready "right now". I have two responses to this: (1) we are making things ready 'right now', just grab any of the available packages and start working and reporting bugs, and (2) there are already 'scaled back' versions of Perl 6 appearing and being used, such as NQP or even the components that execute the standard grammar. Some of these other projects, like NQP, are being used for "real" programs and applications today; they're not simply theoretical exercises or research projects.
Others often claim that all this effort on Perl 6 would be better spent on improving Perl 5. In my experience, those of us working on Perl 6 have absolutely no qualms with seeing Perl 5 improve and continue to grow -- we welcome it. Indeed, many of the features appearing in Perl 5 today come directly from ideas and things first attempted in Perl 6, and we're genuinely happy to see that. But just because Perl 6 developers also like Perl 5 doesn't mean that doing Perl 5 core development is interesting to us, or that (in my case at least) we'd even be qualified to do Perl 5 core changes. We aren't commodity programmers that are interested in simply being unplugged from one project and into some other project that others think is more worthwhile. Personally, I'd prefer to see people who are really into Perl 5 improvements continue to work to make them happen, and that the surrounding ecosystem continue to evolve to enable that to happen more easily for more people. Indeed, this is my wish for all open-source projects, even the ones I don't find interesting or otherwise disagree with.