Perl already enforces bits of the language. E.g., with C++ I can pick to use a regexp library that comes on a UN*X platform, or a C++ regexp library, or (f)lex (for some things), or Henry Spencer's regexp library, or the Perl5 compatible regexp library, or any number of other libraries. What regexp libraries has Perl got? ONE. Can you write another library that will have the same syntax? (Probably) NO. Is the syntax consistent with the rest of the language? NOt really.
Is this a good thing? Probably yes. Perl is still very much a data munging language; it's an important part of being glue. And regular expressions are exceedingly important there, both in terms of expressibility and in terms of efficient manipulation.
More of the same
What containers has Perl got? TWO: arrays and hashes. Want something else? It's not going to be as convenient: no special syntax (stealing syntax from something else is "using other syntax", not having "special syntax"), either less portable (if you use a C or C++ extension) or less efficient (if you extend in Perl), and less familiar to "A Perl Programmer".
Is this a bad thing? Probably not. Perl culture emphasises other things (although Perl6 might improve some things). But sticking to the standards of a language is important: a Perl program, written in the Perl culture (regexps, CGI.pm, Tk, closures, blessed objects, etc.) is understandable to any good "Perl Programmer". Another Perl program, using less standard features, would be harder to understand. E.g. functional Perl is possible, but none too popular, for this reason.
| [reply] |
If you're worried that a standard data structures library will somehow deeply pervert Perl into Java by mandating the use of specific modules, I think you're missing a lot more than just the point of the "SPL" concept.
The reason people get berated about writing their own CGI routines is because they do them badly. Not just slightly bad, not just in ways that are minor and forgivable, but in ways that illustrate what a gigantic hairball this whole CGI thing really is. There are bits of code in CGI.pm that are so stunningly stupid that you would be inclined to remove them, but they are there for a reason, probably to work around some limitation or peculiar "feature" of some Web server or browser somewhere. It is, as best as I can tell, the collected knowledge and wisdom on the subject of CGI in Perl, presented in module format.
Standard data structures are not hairy in this way. Anyone can make a linked list and it will work just fine. You know it is going to work properly because you know exactly what it has to do, and you can test it. To compare, people still aren't sure what CGI is in specific terms.
What educated_foo was probably suggesting was that although many of these are trivial to implement, why should we all have to roll our own each time we need them? When you roll your own, there is always the risk that you will do it wrong, sometimes because of something simple like an off-by-one error, or a mistakenly reversed logical operator. Nobody is perfect, and I would argue that anyone's relative perfection diminishes as the night grows longer.
It would really be great if these sort of structures were built into the standard distribution so that they could be used. That way nobody is forcing you to use them, but at least you aren't forced to write your own.
| [reply] |