httpd-docs mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Antwort: Re: Antwort: [STATUS] (httpd-docs-2.0) Wed Oct 16 23:45 :26 EDT 2002
Date Thu, 17 Oct 2002 21:23:04 GMT


>> Would this be a way that you like the Apache documentation
>> to be available? (This may require some maintenance for
>> itself, to prevent idiots posting rubbish there.)
> Sure, it would be great to have a user-annotation abbility.
> The problem is not finding features that we like.

As the mail was so short about this point I was not sure
at all about this point.

> The problem is finding simple and clean ways to impliment them.
> For example, we really can't load anything on
> unless it is well-tested, very secure, and it comes with an
> active maintainer who is willing to take care of it.

I think finding features that enough people would _agree_
to like might prove less trivial than you believe. ;-)

Having some feature list and some technical restrictions
(like which 3rd-party software it may or must not depend,
on, which language may be used for the source code, whether
it must be Open Source etc.) would be a good start to even
think about whether to create something from scratch or to
use some product for this.

As for something similar to the mySQL annotation system,
I would think of some Apache handler to be applied to some
logical copy (might just be an Alias definition within
httpd.conf) of the Apache docs tree, so that the URLs and
links could stay as they are (you surely don't want the
document formats to change when being used for such an
add-on ;-). And as long as we ignore authentification
(which _is_ part of the mySQL annotation system and may
prove to cause most of the work finally) a first prototype
would have to do not much more than
3. read an Apache docs file whose path name can be derived
   from the PATH_INFO environment variable,
4. read a corresponding annotation file,
5. insert the annotations content into the docs file content
   at the position of the </body> tag,
6. insert a <form> for adding comments at the same position,
7. write the result to stdout.

The form would point back to the same handler script which
then had to
1. append the content of the form to the annotation file
   of this document, and
2. while doing so, synchronize parallel write accesses to
   this file,
plus steps 3-7 listed above, and a little error handling.

As stated above, the real work would then be providing all
the authentication and service stuff, like letting people
register and sending them the password they have forgotten.
These things might then cost a lot of time, depending on
the amount of luxury that would be part of the task speci-
fication - which again would have to be derived from the
features list.
The range of the possible implementations might start at the
level of some customized Guest Book script and end at some
relational database containing the annotations.

But don't take me too serious - I am just brainstorming about
a feature that I am not even sure anyone wants to get - very
similar to the use of CSS in the Apache documentation not long
ago ... ;-)

I just want to check whether the magic 80:20 effect applies
here (that investing 20% of the work might produce 80% of
the result), like in so many cases. But one would need to
know whether 80% would suffice, and of course if there is
any metrics to even measure what 80% would mean at all ...

Regards, Michael

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message