Hi Bernhard:

In message <1357335794.31.0.436770561715.issue2550731@psf.upfronthosting.co.za>
 <1357335794.31.0.436770561715.issue2550731@psf.upfronthosting.co.za>,
Bernhard Reiter writes:
>Bernhard Reiter added the comment:
>I can now see the behaviour your are refering to.

Good, so it's not just me 8-).

(note to people reading this on the tracker, Bernhard has test
environment with a test tracker, sample email meabssages etc that will
be uploaded here at some point).

>However I do not yet fully understand it. Using your email script I get:
>
>SENDMAILDEBUG=1 PYTHONPATH="$PWD" python 
>roundup/scripts/roundup_mailgw.py demo < email_1
>tx_SourceCheckAudit(None) pre db.tx_Source: email, nonce: 
>35cac65328bed1477b21
>tx_SourceCheckAudit(None) post db.tx_Source: email, nonce 
>35cac65328bed1477b21
>tx_SourceCheckAudit(1) pre db.tx_Source: email, nonce: 
>cdc748bdd843bd708bbc
>tx_SourceCheckAudit(1) post db.tx_Source: email, nonce 
>cdc748bdd843bd708bbc

Actually you should see only one pair of audit calls, I am not sure
why there are two pairs. And the first pair looks like it's creating a
new issue not changing issue 1.

>tx_SourceCheckReact(1) pre db.tx_Source: email, recorded: arrived via: email
> 
>cdc748bdd843bd708bbc
>tx_SourceCheckReact(1) post db.tx_Source: email, recorded: arrived via: email 
>cdc748bdd843bd708bbc
>
>The history will only mark the changes done to the "issue" object,
>I wonder if there is a mixup between the values of tx_Source
>for "issue" and "msg" somehow in the way how this is implemented.

Just FYI I didn't implement any tx_Source for the message when I first
saw the journal mixing up changes. I put the tx_Source in the schema
because my plan is to also record the tx_Source for messages so I can
send messages:

  "this arrived via email" (sending via email)

  "this arrived via web"  (sending via the xmlrpc or web interface)

to the ticket and verify the value of tx_Source against the message
contents. If the message says it arrived via email and tx_Source says
email, then having the journal report the message arrived via web
means it's a ournal issue and the code is getting the right tx_Source
to the detectors.

If the detectors are getting the wrong tx_Source info and we are
making security decisions on the basis of that info then the code is
worse than worthless.

>Just an idea. My next step is trying to understand how this all
>is supposed to work.

Yeah that's sort of where I fell down. I tried tracing through it with
the python debugger, but that effectively serialized the transactions
and I didn't see the scrambled info in the journal 8-(.