httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From (Robert S. Thau)
Subject New threaded server snapshot...
Date Mon, 24 Jun 1996 14:27:54 GMT
I've put up another one, same old place:

The most significant new bit included with this release is my fairly
drafty PUT-handler, now in the "support" directory.  This snapshot
also includes code (which really has nothing to do with threading per
se, but these *are* my working sources), to support a notion of
"server child checking" (i.e., was I *really* invoked through the web
server?) for the benefit of potentially dangerous CGI scripts which
would like to trust the web server to do some of their access control;
this machinery is used by the PUT handler.

Also tossed into the mix, since these are my working sources, is a
module for rudimentary PICS support.  

Unfortunately, I'm not sure it would be a good thing to publically
distribute since I'm still waiting for an answer from the PICS gurus
as to whether the thing it does is a close enough approximation to
PICS support for them to want to see it widely distributed.  (All it
does is force the contents of a new <PICS-label> quasi-sectioning
directive out onto the returns from all requests from a given
directory or its subdirectories; it doesn't pay attention to the
PEP-subset Protocol-request: header, nor does it set the PEP-subset
Protocol: response header).

Furthermore, as I discovered at the PICS interoperability tryout I
wrote this for (I'm a grad student, they served free lunch, you figure
it out...), nobody much is supporting PICS headers in server returns
yet (just in HEADs of HTML documents), though Microsoft at least will
strongly consider support for the feature *if* it shows up in a widely
distributed server --- their PICS manager thought it might be a bit of
a headache.  (We could give Microsoft a headache, folks.  Think about
it ;-).

Still, at this point, I'm still waiting for the PICS gurus to say
straight up whether or not it's a Good Thing to distribute at all.  So
even if we agreed that there was nothing wrong with throwing it in as
an optional module technologically or group-politically, we might
have to hold up the release for a bit to get a straight answer...

Other differences from the last snapshot include bug-fixes from the
main line up to the level of the public 1.1b4 release, a bug fix or
two specific to the threaded version, including a cleanup of the Great
Misnaming in the machine-specific bits.

Oh, yes... there are also mods to a couple of the standard modules;
in particular, there are some hacks to mod_cern_meta which (since I
wasn't keeping close track with Andy) are gratuitously incompatible.
Sorry about that... I'll be back in synch in the next one.


View raw message