Re: Patch: EMI UUCP DLRs (final)

Hi,

I can think of 2 cases to handle no destination:

1) There are SMScs that do not include destination in their DLRs (that I 
have found through emails).
2) In the cases where ts is unique, there is no sense in doing a LIKE % 
query for performance reasons.

BR,
Nikos
----- Original Message ----- 
From: "Alexander Malysh" <[email protected]>
To: "Nikos Balkanas" <[email protected]>
Cc: <[email protected]>
Sent: Saturday, June 26, 2010 1:24 PM
Subject: Re: Patch: EMI UUCP DLRs (final)


Hi,

I don't think we have to handle case when no destination given. Destination 
is always there.

Thanks,
Alexander Malysh

Am 26.06.2010 um 10:23 schrieb Nikos Balkanas:

> Hi,
>
> I believe I have accounted for almost all your comments to produce the 
> final working patch. I have no way of testing, other than compilation.
>
> This patch will povide:
>
> 1) Align dlr_oracle.c with the rest. Currently, Oracle does a full dst 
> find/remove/update on each DLR.
> 2) Dst use on find/remove/update is controlled in dlr_find by a single 
> variable, use_dst. This is currently set only at driver level and only in 
> the emi driver, while everyone else has it false. However, very easily, if 
> need arises, it can be set in smsc configuration.
> 3) All DLR handling for all smscs will remain as it used to be till now. 
> Only emi handling changes, hopefully for the better (that is the purpose 
> of the patch :-)).
> 4) It will try to match the last 7 digits of the destination or the length 
> of the destination if it smaller. This is defined in gw/dlr_p.h as 
> MIN_DST_LEN. Didn't want to make it larger, wanted to avoid prefix 
> territory at all costs, since I have seen a lot of mangling there by the 
> SMScs. Besides, I believe that 7 digits give enough resolution
> 5) People using emi, should rebuild their indeces, especially if they are 
> running large batch jobs through EMI. The LIKE % construction is not very 
> efficient.
>
> Enjoy,
> Nikos
> <kannel.diff>