Also, there's really no way to do this safely. Compilation of code can always involve arbitrary execution of code, thanks to BEGIN blocks. Which means the compilation would have to be in a Safe compartment, and there's issues around that. Big issues.

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.

I'm curious. What kind of big issues?

Makeshifts last the longest.

For one, there's no protection against using up tons of memory. Someone could easily write code like "my @foo = 1..1e6; while(1) { push @foo, @foo }" and Safe will allow it.

--MrNobo1024
s]]HrLfbfe|EbBibmv]e|s}w}ciZx^RYhL}e^print

{redo}

It's not 100% safe.

Liz

And when someone posts a Java or Python snippet as a point of comparison with equivalent Perl code? Compile the other language? Or when someone uses <code> tags to format representative script output? Or when someone posts small A vs. B snippets that aren't intended to be compiled as is?

I think this is a problem better solved by human intervention, whether it be the brain of the reader, or the harassment leveled at the writer when necessary.

Strictly speaking (strict HTML, that is) <code> tags should not contain output... Something like <pre> would be more appropriate, although IIRC those tags are deprecated at perlmonks.

Nevertheless, I agree with other monks on this idea. Too risky, and too problematic to be workable.

</ajdelore>

<pre> is not deprecated, but can easily cause scrollbars if you don't pay attention, so their use is discouraged, but not deprecated.

Makeshifts last the longest.

A good idea but not very likely to be implemented by the posters.

Also it would open perlmonks to malicious code.

Finally, as many times the code has some bugs in it (which is the reason the poster turned to perlmonks), a lot of the code will simply refuse to compile/run, either because of syntax errors, logical errors, or "..." somewhere in the code.

CountZero

"If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law

Perhaps strongly nudging posters in to the

use strict;
use diagnostics

way of thinking before posting code could eliminate some of the problems you run in to, let the OP catch some of their own errors, and perhaps help the OP's in the long run as well in understanding the errors.

There is no emoticon for what I'm feeling now.

<code>
BEGIN { do naughty stuff }
</code>

Liz

Another would be that the whole point of most posts, is to try and solve erroneous errors in code, which probably means that the poster *cant* get the code to compile. Forcing them to figure out workarounds just to be able to post would just obscure the problem.

We'd need a whole heap of new tags too, <code>, <code that doesnt compile>, <literal non-code stuff> etc. pp. It really sounds like overdoing it to me.

I suggest you just create a script that downloads code (plenty of examples for that about), and automatically checks it. (And for bonus points, an intelligent syntax-repairer, that removes or comments out '...' etc. ;)

C.

What if you are just putting a snippet of code in? There's lots of times where the code you might put in would be huge in order for it to work, but you know that the problem occurs only in a small section of the code. The code wouldn't really work without the rest, but when you think of it, the rest is just background noise to the real problem.

Keep in mind what most people refer to as "PerlMonks" is just a database of posts & user info selectively converted into html output. You could easily implement your suggestion on your end (albeit a little more difficult) and prevent Perlmonks from having to deal with the security issues.

Thanks for all the responses.

Although many mentioned uses of code that are for snippets, , program output, etc., this could be resolved with a new tag, such as <lit>stuff here</lit>, which would act like code tags in everyway except trying to compile the ontents.

The issues of code that runs in BEGIN blocks is much tougher. This would have to be dealt with or disabled, diminishing the usefulness of this tool.

An additional problem: if perlmonks is on *NIX, (Netcraft indicates FreeBSD) Win32:: modules have no chance of working.

If this were to be implmented at all, it would have to be an advisory, rather than a requirement.

Thanks again, everyone.

Keep in mind that if you prevented me from saying BEGIN {...} that I could always say it another way. Similar tricks can be played with overload as well. Anyone know of other modules I can get to execute arbitrary code during BEGIN?

use lib sub { die 'bad code' };
use foo_bar;
[download]

perl -le 'use if eval("print `echo hello`"), CGI; BEGIN{ print "goodbye"}'

-sauoq
"My two cents aren't worth a dime.";

~~
naChoZ