Finally I am implementing :-D
Randall Stewart <[email protected]> Wed, 17 May 2006 09:35:44 -0400
| Newsgroups | gmane.ietf.rserpool |
|---|---|
| Message-ID | <[email protected]> |
Hi all:
I am finally implementing rserpool again :-D
And I have already came across a few issues I have
with the spec :->
In the common parameters we have-
-----------------------------------------------------------------
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Param Type = 0x6 | Length = 0x8 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Policy=0x4 | Weight |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-------------------------------+
Load: 24 bits (signed integer)
(TBD)
--------------------------------------------------------------
And at least 2 other instances of a 24 bit signed integer...
We obviously were concerned about making a nice 32 bit aligned
parameter when we err.. I wrote this document... but this really
makes no sense.. All machines are either 32 or 64 bit aligned (yes
I remember the 16 and 8 bit aligned ones... but lets not
go back there :-D).. So I propose we change all instances of
24 bit numbers (there may be more I have only gotten to the policy's
in my header file building) into 32 bit numbers.. and if we need
to leave the 3 bytes as "reserved"...
That way we get:
a) 32 bit alignment of the parameter
and
b) Easy, implementation wise, 32 bit access of the actual value..
Of course we waste 3 bytes.. but if we are really worried about
waste then we would not even use a TLV format and hard code
things :-D
Unless I hear objections from y'all.. I will make the above change
for the next pass...
ohh.. thats another thing... I know Thomas as started working on
a socket api.. we probably need to get these structures in that
(if there not already there).. I will have to go dig around and
look after I get online again (I am writting this offline :-D)...
R
--
Randall Stewart
803-345-0369 <or> 815-342-5222(cell)