Mm4 to mm1 …

Mm4 to mm1 … SearchSearch
Author Message
Pierre Habouzit
New member
Username: Pierre_habouzit

Post Number: 3
Registered: 03-2006
Posted on Tuesday, March 07, 2006 - 10:05 am:   

I'm sending mms messages through the MM4 interface, and getting them back through the MM1 one (simulating the sending of an MMS to a local recipient from an other MMSC).

When I submit my MM4_forward.REQ, nowSMS just accept it, and send me a correct MM4_foward.RES message (see the tcp dumps of both communications)

text/plainMM4_forward.REQ
MM4_forward.REQ (35.1 k)

text/plainMM4_forward.RES
MM4_forward.RES (0.6 k)


Though, the message that goes out through the MMSC has nothing to do with the message that I submit: now-sms generated message-id instead of mine, and a default empty content (with a smil and a body.txt reduced to "\r\n").

I attached the relevat part of the mmscdebug log.
text/x-logmmscdebug.log
mmscdebug.log (10.8 k)


sth has to be wrong, but I just miss what.
Pierre Habouzit
New member
Username: Pierre_habouzit

Post Number: 4
Registered: 03-2006
Posted on Tuesday, March 07, 2006 - 11:17 am:   

I've found the problem with my MM4, it was due to some bad encoded headers in the parts headers (which is still surprising, but well…)

Though, I don't understand why NowSMS still modifies the MMS-Message-ID that I carefully set, it seems wrong, as it's supposed to be the same as the mail's Message-ID, that are almost immutable.

What's the rationale behind this ? is that configurable ?

Thanks.
Bryce Norwood - NowSMS Support
Board Administrator
Username: Bryce

Post Number: 5604
Registered: 10-2002
Posted on Tuesday, March 07, 2006 - 04:05 pm:   

Pierre,

Out of curiosity, what was the header problem?

Regarding the MMS-Message-ID, yes, it is changed. However, if a delivery report is requested, the delivery report that is sent back should reflect the original message id.

-bn
Pierre Habouzit
New member
Username: Pierre_habouzit

Post Number: 5
Registered: 03-2006
Posted on Tuesday, March 07, 2006 - 04:32 pm:   

look at MM4_forward.REQ part headers. those are :

[...]
Content-ID: "<VelenS1.amr>"
[...]

because in the WSP header of the corresponding MM1, the Content-ID is quoted, whereas it does not needs to.

I guess that the automatic mm4 to mm1 conversion fails beacuse of that "double" quoting.


As of the Message-ID, I find that a curious choice:
1°) it means that you have to keep all the correspondancies between old messages id's and the new ones
2°) it will defeat any "testing" or benchmark software that try to see if any entering message really goes out on a given interface as it should.

My understanding of a Message-ID (and X-Mms-Message-ID's) is that it's a property of the message that is chosen by the first MMSC the message is submited to, and that it's the only one it should keep in its whole life.

Yes, I saw that the mbox part of your MMS-Id is the relative path to its location on disk, in the queue. But I don't see the need to change the mms-id for that, as the transaction-id you use could hold that information, and you still could keep the original Message-ID, and behave in a more common and expectable manner.

Cheers,
Bryce Norwood - NowSMS Support
Board Administrator
Username: Bryce

Post Number: 5607
Registered: 10-2002
Posted on Tuesday, March 07, 2006 - 06:03 pm:   

Pierre,

That is interesting. I would not have expected quotes around the Content-ID: to cause a problem. I can see that they should not be present ... but I would have expected us to ignore them.

Regarding the Message-ID field, I'd be curious to see what happens with other implementations (but not curious enough to try for myself right now). Our decision to remap the Message-ID is because we use this id for our own message tracking.

And we could not use the transaction-id as a subsitute because the transaction-id is too short lived.

I may be wrong, but I'd expect that other MMSCs might also perform similar message-id remapping.

For testing software, I'd probably use the "Subject:" header for tracking information.

-bn

Pierre Habouzit
New member
Username: Pierre_habouzit

Post Number: 6
Registered: 03-2006
Posted on Tuesday, March 07, 2006 - 06:11 pm:   

removing them from the MM4 I submitted just fixed the problem. So I guess it's the reason.

as of the Subject/Message-ID thing, I think I'll need to do it, even if It's a real pain: notifications do not copy the Subject line (at least it's not mandatory).

honnestly, mail servers do not need to alter the message-id and it seems curious to __have__ to do it.

transaction_id are not that short lived, because nobody enforce you to use a different one each time you re-notify a mobile that an MMS is ready for it.

and you can also use the same transaction_id for different events, since internally you can prefix transaction_id's with the interface through wich you got that transaction_id, making those completely unique. It's a hack, but a nice and safe hack IMHO.