Slashdot Mirror
On Perl 5.6
bryan schwab writes "Perl 5.6 is in development, here is a list of
all the new features" The article is an interview with the "Patch Pumpkin" - one of the
guys from ActiveState in charge of putting together the development release. Seems there is a lot
planned for the next release.
If you were on p5p, I'm quite certain you wouldn't be joining the other Illuminati wannabes with these unsubstantiated delusions of evil conspiracies and take-overs. Please try to simmer down, and do a little research. You'll see that all is well. Or talk to one of us directly if you'd prefer.
0) Tsk tsk on those of you who believe everything Microsoft say. When they said they were paying Activestate to make Perl work better on NT, you forgot to translate that back into honest English: they're paying Activestate to implement fork() on NT so that NT can become more like Unix (a slower buggier more crashprone Unix). Yes, Microsoft for "bring up to baseline of minimum acceptable functionality" is "make it work better". Another way of saying this is "make NT suck less for Perl users". Perl is all about making your OS suck less.
1) Perl's version numbering is changing because major changes in functionality (e.g., the introduction of threads and a compiler) were hidden in the thousands decimal point of the version number. Hopefully this new numbering system will make it more obvious that things have changed. It's a small cosmetic change, not technical, but it's something people wanted.
2) Lexical warnings. Perl's philosophy is "there's more than one way to do it". With warnings as they were, there were only two ways to do it: all warnings enabled, no warnings enabled. Now you can say "I know it looks to the compiler as though I only use this variable once, but don't issue a warning". If you don't see yourself wanting to use it, don't panic--unless you explicitly use the new warning doodads, your programs will emit as few or as many warnings as before.
3) Sarathy is a smart smart guy. He's not an NT flack. He's a smart Unix guy who happens to be employed at Activestate because that's about the only place a guy can be hired to work on the Perl internals. He does far more work on the Perl core to benefit everyone than he does Windoze-specific work. If you're going to be at the Perl+Apache+Linux+... uberconference that O'Reilly are organizing, I urge you to seek him out (he won't be hard to find) and have a talk about your concerns. He'll be the first to agree that NT sucks hard.
4) Unicode reading (via UTF8) is implemented as an add-on module, so that only the part of your program that needs to know about Unicode takes any speed hit. It's also simple to turn programs that used to read ASCII text into programs that read Unicode text: just add "use utf8" at the top. We don't anticipate slowdowns anything like those that one Slashdot reader reported in Tcl.
5) The Unicode support only extends to reading and processing text, not to Perl's interaction with the system (filenames, error messages, etc.) I believe that Windows deals with UTF in the same way that many Unices deal with long files: a separate set of OS functions are provided. The Unicodification of Perl is being extended to use these parallel functions, if present. Generalizing this is hard because, like Unix and long files, everyone does it different. That doesn't mean that Sarathy's work precludes this happening for any other operating systems that provide Unicode support like this. In fact, having the first work done will make support for other operating systems easier. And the best bit is that Microsoft are paying for it.
I don't know about you, but I'm all for fleecing Uncle Bill.
Nat
If it accurately reflected the language, then the site would be pretty ugly. It would, however, probably find a way to bake your bread online and make you a sandwich too; there would be modules for condiments and ...
whoops, shouldn't post when I'm hungry.
"Oh, I hope he doesn't give us halyatchkies," said Heinrich.
As far as casual programmers like myself go, I really appreciate a numbering system that's more "industry standard." I mean, sticking a couple of decimal places out in front of a major revision number is pretty confusing for those of us who can't keep up. Until I read this piece, I'd though 5.005 was only marginally different from 5.004. (Raster, Mandrake: if you're reading this -- why is E still at .16? Are you planning 83 more releases between now and the one where it's finally ready?)
Now, if they'd add support for *irregular* expressions, that's something I could really go for; programmin' Yoda-style!
"Oh, I hope he doesn't give us halyatchkies," said Heinrich.
Certainly there exists a class of problems that lends itself to more than one PC (program counter; it means program counter damn it) executing simultaneously in the same text and data. Personally, I prefer everything unshared unless I say otherwise, not the other way around.
But it's pretty scary stuff for all but the most rigorous of seasoned programmers. To my mind, handing a script kiddie multithreading is like handing a six year old a loaded Uzi. That's not to say that in the hands of a trained professional, these aren't useful. But in the wrong hands, they're more like a sneaky murder-suicide device.
It's getting a bit dated, and certainly carries its own agenda, but several years ago, John Ousterhout gave a talk on Why Threads Are A Bad Idea (for most purposes) that you might like to check out.
If you're curious about threading in Perl, the best place to start is probably the perlthrtut manpage. Some low-level documentation is also available.
I'll be giving a half-day talk on multitasking (both forking processes and spawning threads) in August at the Perl Conference in Monterey. I've got about 100 slides for 3 hours of talk, which should be, um, brisk. :-)