W3C home > Mailing lists > Public > www-validator@w3.org > June 2002

Re: First post; mod_perl and installation issues

From: Terje Bless <link@pobox.com>
Date: Fri, 14 Jun 2002 18:08:45 +0200
To: p kent <pkent77@yahoo.com>
cc: www-validator@w3.org
Message-ID: <r01050300-1015-A5B03198801011D6BB0200039300CF5C@[192.168.1.7]>

p kent <pkent77@yahoo.com> wrote:

>Can I assume that the 0_6_0 branch is stable enough for production use
>of the validator software? I understand that the software's still in
>development though.

0.6.0 is still in mild flux. I consider it alpha ATM. That is to say, I
know there will be more substantive changes in the nature of new features
and changed functionality. I do however expect to get to a beta stage
within weeks -- at which point it should be more or less reasonable to try
using it in production -- barring a few caveats. Once it goes live on
validator.w3.org we consider it stable enough to run in the ultimate
"prodcution environment". :-)


>>see <http://dev.w3.org/cvsweb/validator/?only_with_tag=validator-0_6_0>
>
>Excellent stuff. The one thing that still sticks out in 'check' is the
>use of '&' to prefix subroutine calls. This has the effect of disabling
>prototypes. Some of the subs are defined with prototypes so you're
>losing the benefit of protoypes by using the '&' syntax - so,
>personally, I'd use the normal '$File = handle_uri($q, $File);' syntax
>instead of '$File = &handle_uri($q, $File);'

This won't change / get attention until Validator is refactored into
modules (which it very likely will in one of the next devel cycles, unless
one of those caveats rears it's head); at which time it'll be a moot point
as the majority of them will become methods (but which will certainly have
prototypes etc.).


>I see there's a fair few calls to 'exit', which I'd been told were not
>good under mod_perl (i.e. they caused the entire child process to exit,
>not just the script). Maybe wrap the body of the code in an eval{} and
>use die() to longjump out on a fatal error condition, print error
>message, do cleanup, etc? Or am I wrong about exit()?

Apache::Registry overloads exit() to mean Apache::Registry::exit() intead
of CORE::exit() and works fine under mod_perl (AFAIK).


>Apart from that I wondered why you need to untie *STDIN - AFAICT you
>don't ever read from STDIN (you use CGI.pm, and that's a Good Thing!) so
>I'd like to know how you got bitten by mod_perl - I'm always interested
>in mod_perl gotchas, especially if I can avoid them :-)

mod_perl gets confused when you do certain funky stuff with filedescriptors
and fails to differentiate between STDIN and STDOUT. The recent change to
use secure temporary files triggered this bug.

There is still a bit of cruft left before we declare mod_perl a supported
environment. I probably won't be able to get around to it during the 0.6.0
cycle.



BTW, feel free to stop by #validator on irc.openprojects.net if you would
like to chat about the Validator. If I'm not there, odds are one of the
other guys are. Though we seem to have a large concentration of Europeans
here so the hours may reflect this. There is the odd resident of .jp around
too, but very few .us-anians.

-- 
Editor's note: in the last update,   we noted that Larry Wall would "vomment"
on existing RFCs. Some took that to be a cross between "vomit" and "comment."
We are unsure of whether it was a subconscious slip or a typographical error.
We are also unsure of whether or not to regret the error.      -- use.perl.org
Received on Friday, 14 June 2002 23:33:35 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 25 April 2012 12:14:03 GMT