Folks,

During some testing recently, we discovered an implementation which
barfed if it received a request with a URI whose user name was longer
than some fairly small number - 64 bytes. As a first pass, such a
limitation might seem a reasonable thing to do. After all, usernames are
generally short, right?

Wrong. One of the classic blunders in protocol engineering is building
limits into namespace sizes because of a perceived limitation in use. We
are already seeing discussions of embedding other URLs within the user
portion of the URI; the explosion in web sites is causing domain name
sizes to increase. Expect this trend to continue. Thus, I would STRONGLY
recommend that implementors not introduce artificial limits into the
size of any name related element (user portion, hostname, URL
parameters, branch ID of Via header, etc.). Flexible management of
spaces like these are really important.

Thanks,
Jonathan R.
-- 
Jonathan D. Rosenberg                       72 Eagle Rock Ave.
Chief Scientist                             First Floor
dynamicsoft                                 East Hanover, NJ 07936
[EMAIL PROTECTED]                     FAX:   (973) 952-5050
http://www.cs.columbia.edu/~jdrosen         PHONE: (732) 741-7244
http://www.dynamicsoft.com

Reply via email to