comments about extended log file format

In the proposed extended log file format, a set of identifiers are listed as requiring
a prefix, yet it's not stated clearly which prefixes can be used with
which identifiers.  I think it would be better to say this explicitly.  My interpretation
is:

ip 
     IP address and port, field has type <address>.
     prefixes: "c" (the client making the request), "r" (the proxy chosen to answer the request)
dns 
     DNS name, field has type <name> 
     prefixes: "c" (the client making the request), "r" (the proxy chosen to answer the request)
status 
     Status code, field has type <integer>
     prefixes: "rs" (proxy to server), "sc" (server to client)
comment 
     Comment returned with status code, field has type <text>
     prefixes: "rs", "sc"
method 
     Method, field has type <name> 
     prefixes: "cs", "sr"
uri 
     URI, field has type <uri> 
     prefixes: "cs", "sr"
uri-stem 
     Stem portion alone of URI (omitting query), field has type <uri> 
     prefixes: "cs", "sr"
uri-query 
     Query portion alone of URI, field has type <uri> 
     prefixes: "cs", "sr"

Is this interpretation consistent with the authors' and others' intentions?

I can't seem to think of a reasonable n example where the value of "method", "uri",
"uri-stem" or "uri-query" would differ for "cs" (client to server) or "sr"
(server to proxy).  As far as I can tell, the only value in distinguishing
between these two cases is to indicate just what is logged: if a log file
contains "cs-uri", it logs the URIs which were requested, whereas a log file
which contains "sr-uri", it logs just URIs which were fetched from a higher-level
server.  Yes?

-Adam

Received on Thursday, 25 July 1996 09:58:13 UTC