httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Fielding, Roy" <>
Subject RE: cvs commit: apache-2.0/src/main http_core.c http_protocol.c
Date Mon, 02 Oct 2000 23:03:04 GMT
Yes, I suspect it will need to be "done" by the core filter
(or, to use a better term for it, the message envelope filter)
since that is where chunks need to be coalesced as well.  It
was my impression that Bill's filter would be placed by the
core filter and not by other filters, given that it is static
to http_protocol.c.  That is equivalent to buffering within
the core filter.

Regardless, we used to have a general rule that a veto is only
applicable if the vetoed solution is believed (by the veto wielder)
to be worse than the current distribution package (this was pre-cvs).
I'd rather encourage people to commit better solutions than to sit
around and argue about what has already been committed, even if it
isn't ideal.  Now that we have added filtering code to 2.0, arguments
about design purity must take a take a back seat to getting the code
working in an acceptable manner.  Otherwise we'll never learn
from real experience.

BTW, this will only work if we prevent ourselves from considering
the existing code to be sacred.  IMO, anything within the existing 2.0a
server can and should be replaced by any solution that does "better",
with the exception that we have to keep compatibility in mind
whenever a config file is changed.  That way, we only have to argue
about what "better" means, and not what "pure" means. ;-)


View raw message