However, forcing everybody to add declarations for any runtime-generated subroutines strikes me as both un-perlish and extremely unlikely to happen.
There's no reason it couldn't be a pragma. Why stop
with
use strict; use warnings; use diagnostics; use Taint;
when you can add
use predeclaredsubs; use predefinedsubs;
and maybe
no earlyreturns; no typeglobs;
and possibly
use bondage; use discipline; use pain; use Inline::Pascal;
$;=sub{$/};@;=map{my($a,$b)=($_,$;);$;=sub{$a.$b->()}}
split//,".rekcah lreP rehtona tsuJ";$\=$ ;->();print$/
| [reply]
[d/l]
[select] |
I'm not saying that it cannot be done, just that it might be of limited utility since it would depend on code being written (or re-written) in a way that support the pragma.
The pain of writing code to support the pragma may well be more than the pain saved by the compile time checking.
| [reply] |
The pain of writing code to support the pragma may well be more than the pain saved by the compile time checking.
Oh, I would agree with that. I personally would not
use said pragmas. Then again, I don't even always
use strict (though I do use it in include files and
in large, complex programs, and I would certainly
use it in a module). However, there are some people
whose mindset just tends to prefer a lot of strict
rules. People who prefer Pascal over BASIC, if you
can imagine. People who use strict and warnings even
for one-liners and obfuscations. Those people would
probably appreciate the
ability to force all subs to be predefined, for
example. Of course, if they really want it, one of
them would have to step forward and implement it,
but what I was saying is that if it's done as a
pragma, we needn't object, because all you have to
do to be free of it is not bother to use the pragma.
That said, I'm not sure what the bondage pragma would
do. Perhaps require all control structures to have a
single entry and a single exit point (so, no next or
last permitted, to say nothing of early returns), I
suppose. I imagine the discipline pragma might
prohibit subroutines from accessing any variables
not lexically scoped within the subroutine, so that
any external variables would have to be passed in
as arguments to be used, and if they were to be
changed the results would have to be passed back
out as a return value and applied by the calling code.
I'm not sure what use pain would do, exactly, but I
suppose it would be even worse than discipline.
The really neat thing about these pragmas, if they
existed, would be that we could say, "C gives you
enough rope to hang yourself, but Perl gives you
enough rope to tie yourself up."
$;=sub{$/};@;=map{my($a,$b)=($_,$;);$;=sub{$a.$b->()}}
split//,".rekcah lreP rehtona tsuJ";$\=$ ;->();print$/
| [reply]
[d/l] |
