Unless I misunderstand you, you are referring to the same thing (ie
transport ACK).

Ok - from reading Tom's document (and along the lines of what
Tim is saying), I don't understand how the ACK's that healthlink
is talking about in their "code of practice" are semantically
any stronger than the acks that could be generated by the transport
layer when it receives msgs, or indeed by the clinical application when
it purports to save data into a record. All of these involve a
_computer_ system taking responsibility for the message. Now I
hope we are getting competent computer programmers to
write our programs so I don't know where all these missing
messages are going - it's not like Healthlink is the only company
with the magic formula for using a transactional database.
So if all the ACK is saying is that computer system X has persisted
the message to disk, frankly that is no stronger than doing it at
the transport level boundary.

Now, if instead the ACK is a real acknowledgement that this
path report has been reviewed by a human being and a clinical
judgement formed about the next action to take, that for me is
a real acknowledgement - and that is something that is for me
is best done inside the clinical application. I don't see much value
in pathology review being done in the 'messaging app' because
access to notes, recalls etc are one step away.

In summary, if the ACK they are talking about is a real human
ACK, then I am all for that - but would think it should be done
inside a real clinical application, not inside a glorified ftp client.

Can you indicate what you mean by "integrate fully" and give an example
where the functionality of such "full integration" couldn't be achieved
using the "uncoupled" architecture present in the market?

The current uncoupled architecture gives way too much
importance to 'messaging vendors'. Messaging vendors should
be living in the world of 'database vendors', 'graph component
vendors'. Clinical software companies should pay messaging
vendors for components if they can't be bothered writing their
own component to meet the public standard (identically to the
way a vendor would license an SMTP component if it wanted to enable
email sending in a program). GP's should be purchasing systems
based on the messaging standards they support (not whether
it has Healthlink or Medical-objects as its engine - though the
quality of the licensed messaging engine could be a competition
point much the same way as people currently grumble about
vendors who use SQL server..)

Not any time soon, at least for the transmission of messages between
competing clinical packages. For obvious reasons, clinical vendors "A" and
"B" are more likely to deal with messaging vendor "Z" than each other.

I don't want A and B ever to have met each other - I want them both to
commit to 3rd party approved standard X. I want X to be well defined
enough that it doesn't take a month of developer meetings for A and B
to communicate with each other. If A or B fails to work with a
particular message
(say a difference of opinion in the spec) - I want a third party
arbitrer (AHML) who tells them what the correct behaviour is. IMHO this
is how technical (loosly coupled messaging) standards are meant to work.

Andrew
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk

Reply via email to