To: peterd@bunyip.com
In-Reply-To: Peter Deutsch's message of Fri, 18 Mar 1994 08:29:34 -0800 <9403181629.AA15289@expresso.bunyip.com>
Subject: Re: Unresolved URL issues
From: Larry Masinter <masinter@parc.xerox.com>
Message-Id: <94Mar18.094652pst.2732@golden.parc.xerox.com>
Date: Fri, 18 Mar 1994 09:46:50 PST
Sorry, I was a bit too terse...
I said:
>> Do is your argument against the `multiple CWD' interpretation one of
>> aesthetics or efficiency? I think 'smart clients' could probably
>> recognize a FTP greeting from their favorite style of FTP server, if
>> need be.
And Peter answered:
>I do believe "aesthetics" is a pretty important design
>consideration. Without voting one way or the other in this
>particular debate, I think as a general principle we
>should avoid choices which place too much dependence upon
>"smart clients" as I can't help but see them as examples of
>special casing, which seems to imply our architecture is
>not as clean as it could be.
a) While I think aesthetics are important, they are harder to reach
consensus on, and there are multiple perspectives for aesthetics in
this case. I was just trying to understand on which grounds
Keith was basing his dislike of the 'multiple CWD' choice.
b) Nothing in the `multiple CWD' case requires or depends on
`smart clients'. In fact, perhaps it was a mistake for me
to mention the possibility at all; I was just imagining it
as a possible optimization if you were really worried about
performance; on reflection, I should have merely asserted that
three or four CWD commands in the FTP control session don't
add significantly to any of the costs of setting up a FTP
connection, logging in, establishing the data port, and, finally,
actually retrieving the data.