[Q] Anyone using path info?

From: Terje Bless (link@tss.no)
Date: Sat, Sep 04 1999


Message-Id: <199909041200.OAA06024@vals.intramed.rito.no>
Date: Sat,  4 Sep 1999 14:05:57 +0200
From: Terje Bless <link@tss.no>
To: W3C Validator <www-validator@w3.org>
Subject: [Q] Anyone using path info?


I have a patch sitting ready that will make the validator use the CGI
module (which, among other things, will make file upload trivial to
implement). Using CGI rather then rolling your own CGI routines is also
much more efficient and makes for a far cleaner implementation.

However, to remain bugwards compatible with the existing validator, I had
to keep most of the old CGI code to deal with the cruft that /might/ be in
the extra path info part of the request. That is not a good situation. It
makes the code rather hairy and it partly defeats the purpose of using
CGI.pm.

Does anyone actually use this feature?

Perhaps Gerald can do some fancy grepping on the logs to see if there are a
significant number of requests that contain path info?

IMO this is a misfeature in any case and should be removed, but now it's
holding back further development of the code, which makes it even worse. Of
course, my priorities may be wildly different from Gerald/W3C's in this
regard, but I can't see that this "behaviour" needs to be there; at least
if you put up a "no longer supported" page for these requests in a
transition phase (the "check/referer" bit can still be retained!).


I would appreciate people's opinions on this. Want to chime in here Gerald?



The reason I ask is that I am, for the second time, in the process of
developing new features for the validator for use on an intranet that's
behind a firewall. The experience from the first time round shows that it's
not feasible to maintain local diffs and merge in Gerald's changes. Unless
I get patches accepted into the official source base, I'll have to
completely fork the code such that submitting any of the code back to the
W3C becomes impractical (which is a serious lose-lose situation in that I
don't get access to the official changes from Gerald and I can't submit
patches back to him).

I *really* want to avoid that! It happened last time and the code is now
impossible to release because my then employer holds the rights to those
modifications (which means I'll have to start from scratch this time
around. *sigh*).

My priorities are roughly consistent with the Validator's TODO list, but
with the added priority of cleaning up the code some and modernizing it (it
practically _reeks_ of Perl 4). A side effect of the last priority is that
the code becomes *much* easier to maintain, less memory hungry, less prone
to race conditions and similar bugs (temp-file handling is sub-optimal) and
quite possibly faster. Unless Gerald has issues with my implementation
details, our priorities should coincide or at least complement each other
(i.e. I'm aiming to make this a win-win situation).


Anyway, this CGI path info situation is somewhat of a show stopper for me.
It's hardly impossible for me to keep this code around, but it's rather
painful to muck about with this and it's horribly time consuming. I'd like
to hear back from this list in general, and Gerald in particular, before I
make a call on this one. If I can, I'll make a patch that strips the path
info gunk (which, BTW, doesn't work in the official validator!) and submit
that to the list; otherwise I'll send the patch as it is today and try to
keep it around as long as possible as I add features.


TIA for all comments and opinions. TTFN, -link.

-- 
*** I just switched to a new email client.
*** If you see any format problems in this message, yell. Loudly! :-)

                                             -link