Re: Protocol Draft - PathDataTLV

Jamal Hadi Salim <> Tue, 01 April 2008 19:48 UTC

Message-Id: <TUE.1.APR.2008.154825.0400.>
Date: Tue, 1 Apr 2008 15:48:25 -0400
From: Jamal Hadi Salim <>
Organization: ZNYX Networks
Subject: Re: Protocol Draft - PathDataTLV
Comments: To: Evangelos Haleplidis <>
Content-Type: text/plain
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit

Greetings Evangelos,

Much thanks for the review; comments below.

On Mon, 2008-31-03 at 22:56 +0300, Evangelos Haleplidis wrote:


> 1. Path-Data-TLV is the only TLV (besides KEYINFO-TLV) that is not depicted
> as a graph in the document. While it is described, it would be good for
> consistency to be seen. In the end of the document I include an example.

Looks usable to me. If there are no objections from the other authors,
we should include it.

> 2. In the second text the part "when PATH flags are 00" should be changed to
> "when PATH flags are 0x00", unless there was intended as 0b00. It should be
> specified however for clarity.

I think that paragraph is a bit confusing; please refer to the other
text I am suggesting below to replace that section.

> 3. About Selector Bit, in which position does it exist in the flags? I guess
> it should be the first. However I think it should be documented for clarity
> purposes (I also include a graph in the end of the mail).

Hrm. Good catch. Our implementation has it at the other end. i.e. bit 15
May i suggest we put it there?

> 4. I don't understand exactly what the text in page 48 regarding FullData
> TLVs mean. If PathFlags!=0x00 then the FullData MUST contain inside the V
> the row besides the value, the index of the row (fifth row for example)? But
> if PathFlags==0x00 then it MAY contain that OR the row will be specified in
> the Path (IDs)? While I searched I did not see an example in the document,
> or perhaps I was searching wrong.

This is the same in concern of #2. So for both #2 and #4 here's text
which i think provides more clarity.

When a table is referred to in the PATH (IDs) of a PATH-DATA-TLV,
then the FULLDATA's "V" will contain all of that table's row contents.
Refer to blah for an example of this.
It is possible to reference a specific single row by appending
to the PATH (IDs) the index of the specific table row. For an
example of this refer to blah. If a a table row is referenced
in this way then in order to avoid ambiguity, the PATH flags must
MUST ensure that the content selection flag (F_SELKEY) is not set.
Setting F_SELKEY will result in blah error.

I will look up the examples and replace the blahs.
By not specifying the 00 leaves us room in the future should we add more
flags to be make sure those are not zeroed out when needed..

Does this provide more clarity?

Thanks again for the scrutiny.