From: Noveck, Dave (Dave.Noveck@netapp.com)
Date: 09/21/01-08:44:46 AM Z
Message-ID: <8C610D86AF6CD4119C9800B0D0499E3333561C@red.nane.netapp.com> From: "Noveck, Dave" <Dave.Noveck@netapp.com> Subject: RE: Issues list for RFC3010 Date: Fri, 21 Sep 2001 06:44:46 -0700 Spencer Shepler wrote: > On Fri, Peter ?strand wrote: > > > > > I vote to remove "tag" at all. If it used for debugging - use snoop, if it > > > > > used for caching - use something else (xids from RPC) > > ... > > > I like the idea of a fixed length for the tag. > > > > > > It seems that its variable length is more of a > > > problem than its presence. > > > > Why is the variable length a problem? > > > > My opinion is that the tags should be included in the protocol, in one > > form or another. They are really useful, IMHO. > > Variable length adds overhead to the decoding of the request. Not that much. A fixed length would mean that all clients would have to bear additional network overhead even if they didn't want to use the feature. A variable-length allows the client to choose a length of zero and only pay four bytes. My first choice is no tag. My second choice would be a variable-length tag with the server encouraged to return the tag unmodified in the response but allowed to truncate it because of resource issues caused by long tags.
This archive was generated by hypermail 2.1.2 : 03/04/05-01:49:08 AM Z CST