The recent discussion that started with Ovid’s Perl 7 blog post has me thinking about the future of Perl, and Perl 5 in particular. We hear that Perl is dying on a regular basis, and while I take that with a grain of salt, the fact that so many non-Perl people seem to believe this is worrisome. If Perl is to have a future, it will need to attract new users.
Perl 5, as it stands, has a lot of problems. One of them is that people outside the community don’t want to learn a language that’s soon going to be replaced by Perl 6. While Larry has said repeatedly that Perl 5 and 6 are two different languages in the same family, and that Perl 6 is not intended to replace Perl 5, very few people outside the echo chamber have heard this message.
Another problem is that Perl 5 has a lot of baggage. It drags that baggage around in the name of backwards compatibility. Perl 5’s dogged commitment to backwards compatibility is, on the whole, a good thing. However, when combined with Larry’s proclamation that there will never be a successor to Perl 5, we have a truly serious issue.
If it’s true that Perl 5 and 6 are really two separate languages, then we need to be able to evolve Perl 5 and 6 separately. But how can Perl 5 evolve when there’s no way to develop a successor to Perl 5? Perl 6 is not the successor, it’s just a sibling.
Perl 5 is stuck with an increasingly crufty core implementation tied to an increasingly crufty set of no-longer-correct decisions. Some of those decisions include smartmatch, half-assed overloading, no strict by default, indirect method call syntax,
-> instead of
. for method calls, half-assed exceptions, and many more.
It’s unlikely that there will ever be a Perl 5 release which fixes all of these problems, because there’s no way to fix them without breaking back-compat. Without the option of bumping the major version number, how can we signal that any new version is substantially different than the last release?
Realistically, the only way to evolve Perl 5 is going to be to create a new core that implements a Perl 5-like language. Stevan Little’s Moe is one stab at a possible successor, though don’t get too excited, for now it’s just an experiment.
The future of Perl 5 is, sadly, not going to be called Perl. So if you’re interested in seeing a future for Perl 5, the discussion about Perl 7 (and 6) is a distraction. Instead, figure out how to create a new Perl 5-like language. Maybe you can contribute to Moe. Maybe you can start your own Perl5++ prototype project. But you can’t convince p5p to call perl 5.20 Perl 7, and you can’t rename Perl 6.
I’m sad that it’s come to this. Perl 6 should have been renamed years ago to preserve the health of the Perl 5 community. I’ve seen many people say that this is solely Larry’s decision, but that’s just wrong. A lot of people have invested a lot of time and effort in Perl 5, and those people count too. I don’t think their interests have been taken into account.
Of course, I look forward to being proved entirely wrong when a production-ready Perl 6 is released this Christmas and the entire Perl community moves as one to embrace it.