httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From (Geoff White)
Subject Re: Survey information...and names for Apache
Date Tue, 05 Dec 1995 00:14:55 GMT
At 10:31 AM 12/1/95 -0800, wrote:
>My opinions:
>1) Sameer (and Ben Laurie) should make it priority #1 to identify how to
>   turn SSL capability (and SHTTP as well - Geoff, you want to join in 
>   with them?) into a module that can be released independently.  Full
>   integration with us in the US collaborating is just running too many 
>   risks to be worth it, and the API needs expanding at any rate.
>2) Sameer should use the word "Apache" in the name of his software.
>	Brian
>  http://www.[hyperreal,organic].com/

The server that I am completing right now is based on Ben's work reved up
to 8.6.16.  It uses Terresa's Secure Web toolkit to do SHTTP and SSL,I use
very little of ben's code but I do follow his "os_conn" strategy as I
thought it was
the best way to head towards a "protocol API" which my code falls far short
of being. The current server that I'm debugging as we speak just has the SWT
calls "wedged in", at first I though that I'd have time to do a proper job
but this has to be a "quickie" as I'm under some amount of delivery pressure:

        The correct way to do this is to spec and build a protocol API.
        SSL (and SHTTP even more so) can't be implemented as basic Apache
        modules (I found that out fairly early) What would work is if Apache
        had a Protocol API as a seperate or enhanced version of the existing
        module API then SSL and SHTTP could be plug-ins.  One could even
        do HTTP-NG this way, but certain aspects of HTTP-NG are more happier 
        if you have native multi-threading.

        Also, a Protocol API would not be a "hook for encryption" it would
        be a way to add other useful protocols like Z39.50 [sic] and
        newer experimental versions of existing protocols.

        The server that I would LIKE to build/participate in building would
        have the following features:

        o       Apache 1.0 base
        o       Enhance and clean-up the "Module API"
        o       Create a New "Protocol API"
        o       Server core modifications to make the code MT safe
        o       Extra work on the DLD module

        If we had these features then I believe the following would happen:

        1) We could have plug-and-play compatibility for free and "commercial"
           security technology (Ben's SSL work and Teressa's SWT), commercial
           add-ons could be sold as binaries that are dynamically loaded thus
           protecting their intellectual property, and keeping Apache free.
           all of the changes to Apache outlined above could be freely given 
           back to the Apache effort.
        2) A good platform from which people could do affordable HTTP-NG
           and other types of Web research.

        3) A solid platform for various other server technologies that
           need to be developed.  We need to prototype some new things
           for future projects here, the current Apache is good but not quite
           good enough to seemlessly do protocol reasearch, a Protocol
           API would be a good solution.

        4) A viable alternative to commercial servers that might have some
           other corporation's or government's agenda at stake. Some of the
           recent news reports give me the willies :)

This is the project that I would like to work on next, whether it will be
here at EIT or elsewhere is still up in the air at this point, but I have got
positive feed back from a lot of parties.

Please don't think that by any means I even think that I can accomplish all of
this on my own without at least input from everyone here, the work that I've
been doing over the last couple of months has only helped in getting me
enmeshed in the Apache code, I am, by no means an expert in any of these
technologies, (except maybe general un*x daemon technology) but I do feel
that I can contribute to this type of effort and will plod along even if no-one
else is interested.


Comments? Flames?

View raw message