Re: cvs commit: apache-1.3/src/main http_config.c http_protocol.

["Roy T. Fielding" <fi...@kiwi.ics.uci.edu>]

>If you are doing something as silly as limiting the request body size, you
>sure do need to let the user configure it.

What, exactly, is silly about limiting the request body size?

....Roy

Re: cvs commit: apache-1.3/src/main http_config.c http_protocol.

[Marc Slemko <ma...@worldgate.com>]

On Sun, 9 Aug 1998, Roy T. Fielding wrote:

> >If you are doing something as silly as limiting the request body size, you
> >sure do need to let the user configure it.
> 
> What, exactly, is silly about limiting the request body size?

Because if someone uses request bodies for something useful then having a
fixed, unconfigurable limit at something like 32 megs is horribly broken.

Because, while the number of header fields, size of them, etc. are
generally reasonably independent (given currently defined HTTP headers and
use of them) of what you are doing, the request body size can vary a huge
amount.  Something as simple as using PUT to send a 40 meg file (which
seems like a resaonable thing to me) would run afoul of this limit.  It is
only mildly annoying if it is configurable, but if it is a fixed
compile-time thing as was suggested, then it becomes lame.

Oh, why were the DEFAULT_LIMIT_* defines added to httpd.h _after_ the
section where things like that are stored instead of with all the other
config type constants?