From mboxrd@z Thu Jan 1 00:00:00 1970 MIME-Version: 1.0 In-Reply-To: References: Date: Fri, 15 Oct 2010 07:57:19 -0700 Message-ID: From: David Leimbach To: Fans of the OS Plan 9 from Bell Labs <9fans@9fans.net> Content-Type: multipart/alternative; boundary=001636c927f00e7be90492a90bf6 Subject: Re: [9fans] =?iso-8859-7?q?=F0p?= Topicbox-Message-UUID: 65d456ea-ead6-11e9-9d60-3106f5b1d025 --001636c927f00e7be90492a90bf6 Content-Type: text/plain; charset=ISO-8859-7 Content-Transfer-Encoding: quoted-printable 2010/10/15 Sape Mullender > Are we talking about =F0P or 9P? > It's about both. I was just curious about how 9P was deficient in terms of pipelining. It might not be optimal for all cases of pipelining, but the protocol seems to support it in certain cases just fine. =D0P deals with it in a superior way, and I need to finish reading the pape= r on it. > > =D0P doesn't have Twalk. It has open, which combines clone, walk, and op= en > from > 9P. Before you start jumping up and down and telling me that you can't > open > without revieing the qids from the walk (to check them for mount points), > let > me tell you that we are also tackling mount tables. Mount tables will no > longer > match qids but longest prefix path names. > We know the semantics are different. But you have to look hard to find > realistic situations where the difference matters. > > I intend to write a =F0P design document that explains the whole concept = in > excruciating detail. There's a lot more to it than just changing walk an= d > open. > I'm looking forward to it! > > Sape > > > > From: lucho@ionkov.net > > To: 9fans@9fans.net > > Reply-To: 9fans@9fans.net > > Date: Thu Oct 14 23:13:59 CES 2010 > > Subject: Re: [9fans] =F0p > > > > It can't be dealt with the current protocol. It doesn't guarantee that > > Topen will be executed once Twalk is done. So can get Rerrors even if > > Twalk succeeds. > > > > 2010/10/13 Venkatesh Srinivas : > > >> 2) you can't pipeline requests if the result of one request depends = on > the > > >> result of a previous. for instance: walk to file, open it, read it, > close > > >> it. > > >> if the first operation fails, then subsequent operations will be > invalid. > > > > > > Given careful allocation of FIDs by a client, that can be dealt with = - > > > operations on an invalid FID just get RErrors. > > > > > > -- vs > > > > > --001636c927f00e7be90492a90bf6 Content-Type: text/html; charset=ISO-8859-7 Content-Transfer-Encoding: quoted-printable

2010/10/15 Sape Mullender <sape@plan9.bell-labs.co= m>
Are we talking about =F0P or 9P?

It'= ;s about both. =A0I was just curious about how 9P was deficient in terms of= pipelining. =A0It might not be optimal for all cases of pipelining, but th= e protocol seems to support it in certain cases just fine.

=D0P deals with it in a superior way, and I need to fin= ish reading the paper on it.
=A0

=D0P doesn't have Twalk. =A0It has open, which combines clone, walk, an= d open from
9P. =A0Before you start jumping up and down and telling me that you can'= ;t open
without revieing the qids from the walk (to check them for mount points), l= et
me tell you that we are also tackling mount tables. =A0Mount tables will no= longer
match qids but longest prefix path names.
We know the semantics are different. =A0But you have to look hard to find realistic situations where the difference matters.

I intend to write a =F0P design document that explains the whole concept in=
excruciating detail. =A0There's a lot more to it than just changing wal= k and
open.

I'm looking forward to it!
=A0

=A0 =A0 =A0 =A0Sape


> From: lucho@ionkov.net
> To: 9fans@9fans.net
> Reply-To: 9fans@9fans.net
> Date: Thu Oct 14 23:13:59 CES 2010
> Subject: Re: [9fans] =F0p
>
> It can't be dealt with the current protocol. It doesn't guaran= tee that
> Topen will be executed once Twalk is done. So can get Rerrors even if<= br> > Twalk succeeds.
>
> 2010/10/13 Venkatesh Srinivas <me= @acm.jhu.edu>:
> >> 2) you can't pipeline requests if the result of one reque= st depends on the
> >> result of a previous. for instance: walk to file, open it, re= ad it, close
> >> it.
> >> if the first operation fails, then subsequent operations will= be invalid.
> >
> > Given careful allocation of FIDs by a client, that can be dealt w= ith -
> > operations on an invalid FID just get RErrors.
> >
> > -- vs
> >


--001636c927f00e7be90492a90bf6--