RE: [protocol] maxInt

[John McMeeking <jmcmeek@us.ibm.com>]

I agree with keeping the existing limits, though I hope an implementation
would return protocol error if presented with a value greater than 2^^31 -
1 for these fields.

messageID - message ids are assigned by the client.  Increasing the current
limit (for any of the fields) would require a mechanism for the client to
determine that the larger range is accepted by the server -- protocol
version = 4 or something like that.

timelimit - I doubt there is any need for timelimit to exceed the current
limit of 68 years ;-)

sizelimit - I have a hard time seeing a need for specifying a search limit
greater than 2 billion entries.  If this limit were number of bytes, rather
than number of entries, maybe.


John  McMeeking


owner-ietf-ldapbis@OpenLDAP.org wrote on 09/26/2004 10:44:54 PM:

> Well ... no. You've pointed out that maxInt is architecture
> dependent, but for interoperability I think the value can't depend
> on architecture.
>
> If you send the messageID 2^^32 + 15 to a 32-bit system, you could
> reasonably expect to receive the messageID 15 in response, which
> would naturally be rejected.
>
>
> It is a tricky problem, but keeping the smaller number is the best
> bet for interoperability.


>
> Ron
>
> -----Original Message-----
> From: owner-ietf-ldapbis@OpenLDAP.org
> [mailto:owner-ietf-ldapbis@OpenLDAP.org]On Behalf Of Vithalprasad
> Gaitonde
> Sent: Saturday, 25 September 2004 00:47
> To: ietf-ldapbis@OpenLDAP.org
> Subject: [protocol] maxInt
>
>
> The [protocol] draft defined maxInt as (2^^31 - 1). Does it make sense
> to change this to (2^^63 - 1) to make use of 64 bit platforms. This will
> affect messageID, sizeLimit and timeLimit.
>
> Thanks,
> Prasad
>
>