Hi Alex,

patch look good and if no objections from others I will commit it to meta-data branch.

As for the message_id. It would be great to have it in meta-data but if I think a bit more about it, it will become very messy.

How about defining new field in Msg struct (what you did already) and name it like foreign_id. This field will be logged to access log and can take message id from any SMSC module?

Thanks,
Alex

Alejandro Guerrieri schrieb:
Alex,

Done. Please check the attached patch.

There's still the issue regarding the "message_id" param on submit_sm_resp / data_sm_resp.

The new match by tag method voids the option of using the same filtering mechanism mentioned on my previous, mail since it would require a match by name (or inventing a tag
address for message_id, which is not an option IMHO).

Since it's not a TLV, does it make sense to add it to meta-data, or maybe exploring other ways to be able to retrieve it? Maybe a special % param or a conf directive like "smsc-message-id-into-meta-data" on the smpp group. Do you have any other ideas about how to achieve this?

Regards,
--
Alejandro Guerrieri
[EMAIL PROTECTED]



El 26/11/2008, a las 06:43 a.m., Alexander Malysh escribió:

Hi Alex,

great!!! :)

Could you please change this patch to use dictionary tlv_by_tag because the name of configured TLV may be different of the one in SMPP spec. but the tag will be equal. I think it's up to user which name configured TLV should use.

And minor issue: please always make patches from gateway root directory with cvs diff -Nau.

Thanks,
Alex

Alejandro Guerrieri schrieb:
Hi,
This patch allows meta-data to carry all available TLV's not only the User-Defined. All TLV's defined on smpp_pdu.def were "hijacked" by the main pdu structure, so they never reached the tlv dictionary. What this patch does is to check for defined TLV's on the smpp-tlv group and copy them to the tlv dictionary. Those TLV's are then available on the meta-data parameter. It only copies the TLV's explicitly defined, otherwise the meta-data parameter would be unnecessary cluttered with all available TLV's. This solves the "receipted_message_id" issue with deliver_sm (to name one), but does _not_ solve the "message_id" param on submit/deliver/data_sm_response, since message_id is not a TLV. For that parameter I could add a call for meta_data_set_value to inject it into the meta data (already tried and works), but then it would be always available. To avoid this, I could use the same filtering mechanism as with the TLV's, but that would mean defining a dummy tag address, since this is not a TLV so it doesn't have a documented address. I could filter using tag_by_name, so the address wouldn't matter anyways, but it's somewhat ugly imho.
Ideas? Opinions?
Regards,
--
Alejandro Guerrieri
[EMAIL PROTECTED]





Reply via email to