Re: [ftpext] Fwd: New Version Notification for draft-bryan-ftpext-hash-00.txt

[Mat Berchtold <[e-mail]>]

> It can't, at least not like this, simply because if the server gets, e.g.:
It depends how pathname is defined. If it's imported from Section 2.2 of [RFC3659] as currently proposed I agree with you. But I'm also okay with the mandatory start/stop points:
>hash-command = "HASH" SP start-point SP end-point SP pathname

The main concern I have is with the dependency on RANG in HASH. 

-----Original Message-----
From: Sob [mailto:[e-mail]] 
Sent: Saturday, 07 April, 2012 04:33
To: Mat Berchtold; Anthony Bryan
Cc: [e-mail]
Subject: Re: [ftpext] Fwd: New Version Notification for draft-bryan-ftpext-hash-00.txt

At 03:10 7.4.2012, Mat Berchtold wrote:
>The range can be passed directly to the command as optional arguments:
>hash-command = "HASH" SP pathname [ SP start-point [ SP end-point ] ]

It can't, at least not like this, simply because if the server gets, e.g.:

HASH somepath 10 20

.. it can mean any of the following and it's impossible to tell which one:

a) pathname = "somepath", start-point = 10, end-point = 20
b) pathname = "somepath 10", start-point = 20
c) pathname = "somepath 10 20"

It would work with reversed order, with mandatory start and end points before pathname, but if RANG catches up (and it should, because it has big potential), wouldn't it be a little weird to have separate way for using ranges with HASH?

--