From: Jim Rees (rees@umich.edu)
Date: 08/08/02-06:30:13 AM Z
Subject: readdir and eof
From: Jim Rees <rees@umich.edu>
Date: Thu, 08 Aug 2002 07:30:13 -0400
Message-Id: <20020808113013.D5466207C1@citi.umich.edu>
I know no one wants to extend the issues list at this point, but I would
just like to observe that I had a very hard time implementing readdir in my
client from the spec. In particular, there is no discussion of what the eof
flag means.
Yesterday, for example, I came across a server that never sets the eof flag.
Instead it signals eof by returning no entries. We decided that the server
was wrong in this case, but you wouldn't know that by reading the spec.
I suggest we lift the last setence from the rfc1094 readdir description and
make it the last sentence of our DESCRIPTION too:
The "eof" flag has a value of TRUE if there are no more entries in the
directory.
There is also a possible ambiguity in the spec. Near the top of page 185 we
have
For some file system environments, the directory entries "." and ".."
have special meaning... they should not be returned to the client...
But then at the top of page 186 we have
Since some servers will not be returning "." and ".." entries...
I suggest we change "should" to "must" in the first reference, and remove
the word "some" from the second.
One more thing. What error does the server return if the cookie verifier is
no good? Shouldn't this be specified so that the client has some hope of
recovering?
This archive was generated by hypermail 2.1.2 : 03/04/05-01:50:09 AM Z CST