| Author |
Message
|
| angka |
 Posted: Thu Sep 14, 2006 12:27 am Post subject: |
 |
|
Chevalier
Joined: 20 Sep 2005
Posts: 406
|
Hi,
My input message is text(no format). I didnt set any message domain in the MQInput node and broker will parse my message body to BLOB and so my InputBody is in BLOB. so do i SET OutputRoot.BLOB = InputBody;? I try setting this, but when the message leave the Compute node, the MQRFH2.mcd.Msd become blank..
Actually my MQ server is supposed to receive messages in text(MQSTR) to Queue A and then the MQInput node will pick up from Queue A and do some computation to determine what Message Type it belongs to. After adding in the MQRFH2 header(setting the message domain, message set, etc), the MQOutput node will put the message to Queue B. Then another MQInput node will pick up from queue B and mapped it to XML.
Thanks |
|
| Back to top |
|
 |
| kimbert |
| Posted: Thu Sep 14, 2006 12:59 am Post subject: |
|
|
Jedi Council
Joined: 29 Jul 2003
Posts: 5543
Location: Southampton
|
If this is your requirement...
| Quote: |
| Actually my MQ server is supposed to receive messages in text(MQSTR) to Queue A and then the MQInput node will pick up from Queue A and do some computation to determine what Message Type it belongs to. After adding in the MQRFH2 header, the MQOutput node will put the message to Queue B. Then another MQInput node will pick up from queue B and mapped it to XML. |
...then I think there's a much simpler way to achieve it. You do not need two message flows, and you do not need to construct an MQRFH2 header anywhere in your message flow.
MQInput - Domain = BLOB
Compute - Cast input as char, analyse the bitstream to determine the message type, use CREATE FIELD...PARSE with the DOMAIN clause and supply the appropriate message type.
That will get the message parsed correctly. Serializing as XML should be trivial after that. I presume you are using the MRM TDS format to parse the text? |
|
| Back to top |
|
|
| angka |
| Posted: Thu Sep 14, 2006 1:40 am Post subject: |
|
|
Chevalier
Joined: 20 Sep 2005
Posts: 406
|
Hi,
kimbert: Think this may work for me because there are many different type of input text. it may increase in the future. and if there is a new type i will need to add lint to the code. i am using the model tt i mention because if there is a new type i just need to add on to my .mxsd definition. I am using the Custom Wired Format. Thanks |
|
| Back to top |
|
|
| supreeth |
| Posted: Thu Sep 14, 2006 2:32 am Post subject: |
|
|
Voyager
Joined: 17 May 2005
Posts: 90
Location: London
|
Hi,
From the code i have put in my earlier mail, replace
OutputRoot.XML = InputBody with
OutputRoot.BLOB.BLOB = InputBody.BLOB;
cheerz!!!
supreeth
_________________
Supreeth Gururaj
IBM Certified WMQ Solution Expert |
|
| Back to top |
|
|
| angka |
| Posted: Thu Sep 14, 2006 2:43 am Post subject: |
|
|
Chevalier
Joined: 20 Sep 2005
Posts: 406
|
Hi,
I tried, it is still the same. the OutputRoot.MQRFH2.mcd.Msd become none. this looks simple but dunno why i am stuck so long. Thanks.. |
|
| Back to top |
|
|
| kimbert |
| Posted: Thu Sep 14, 2006 3:01 am Post subject: |
|
|
Jedi Council
Joined: 29 Jul 2003
Posts: 5543
Location: Southampton
|
| Quote: |
| i am using the model tt i mention because if there is a new type i just need to add on to my .mxsd definition |
Fair point. I agree that maintainability is important here. All the same, I think there must be a better way.
Have you looked at the embedded message support in v6? You can identify the body of a message using one of the fields in the header. Search for 'Message Identity' in the docs. If you use this technique, you get maintainability AND performance. |
|
| Back to top |
|
|
| supreeth |
| Posted: Thu Sep 14, 2006 3:19 am Post subject: |
|
|
Voyager
Joined: 17 May 2005
Posts: 90
Location: London
|
yups angka ... u r right ... looks like a bug. Theres a stray tag created which is named Format in the same level as mcd. Guess some problem with using BLOB. Try raising a PMR. I guess its an eligible issue to raise PMR 
cheerz!!!
supreeth
_________________
Supreeth Gururaj
IBM Certified WMQ Solution Expert |
|
| Back to top |
|
|
| supreeth |
| Posted: Thu Sep 14, 2006 9:30 pm Post subject: |
|
|
Voyager
Joined: 17 May 2005
Posts: 90
Location: London
|
hey,
its logical. u cant parse the incoming message as BLOB or XML and set message details in MQRFH2. It would be reset accordingly. Should have understood it earlier 
what you can probably do is redesign your message flow. I dont see any need for you to have 2 message flows.
kimbert's idea makes perfect sense here.
| Quote: |
MQInput - Domain = BLOB
Compute - Cast input as char, analyse the bitstream to determine the message type, use CREATE FIELD...PARSE with the DOMAIN clause and supply the appropriate message type.
That will get the message parsed correctly. Serializing as XML should be trivial after that. I presume you are using the MRM TDS format to parse the text?
|
| Quote: |
Think this may work for me because there are many different type of input text. it may increase in the future. and if there is a new type i will need to add lint to the code. i am using the model tt i mention because if there is a new type i just need to add on to my .mxsd definition. I am using the Custom Wired Format.
|
couldn quite get as what issues you would have. i guess using the esql the details can be set dynamically. i mean ... cudn quite get ur problem 
I guess you can just go ahead with the above design.t would be generic to suit your needs and also would work without an issue 
cheerz!!!
supreeth
_________________
Supreeth Gururaj
IBM Certified WMQ Solution Expert |
|
| Back to top |
|
|
| angka |
| Posted: Fri Sep 15, 2006 1:32 am Post subject: |
|
|
Chevalier
Joined: 20 Sep 2005
Posts: 406
|
Hi,
Is there a way not to parse the incoming data message as any domain (remain as text) and then set the message details in MQRFH2?
MQRFH2 isn't it a header for the MQInput node to parse the message accordingly? when the message is put to a queue all the logical tree will be gone also. I used rhfutil it can work even when the message body is not in the format of the MQRFH2 setting. Why did it reset the setting?
Btw, I have already submitted a PMR. =) IBM going to shoot me. Thanks |
|
| Back to top |
|
|
| angka |
| Posted: Fri Sep 15, 2006 1:38 am Post subject: |
|
|
Chevalier
Joined: 20 Sep 2005
Posts: 406
|
Hi,
kimbert: I have not look at that yet. Are you referring MQMD message id as the "message identity"?
Actually All the system that is going to connect to my server is already running live now so it is not possible for them to include anything more.
Thanks. |
|
| Back to top |
|
|
| kimbert |
| Posted: Fri Sep 15, 2006 2:05 am Post subject: |
|
|
Jedi Council
Joined: 29 Jul 2003
Posts: 5543
Location: Southampton
|
| Quote: |
| Is there a way not to parse the incoming data message as any domain (remain as text) |
If you don't want to parse the message, use the BLOB domain.
| Quote: |
| and then set the message details in MQRFH2? |
In your case, I see no reason to add an RFH2 header in the middle of a message flow. See my previous post
| Quote: |
| MQRFH2 isn't it a header for the MQInput node to parse the message accordingly? |
If an incoming message has an MQRFH2 header with an mcd folder, the message properties will be set from its Domain/Set/Type/Fmt fields, overriding any settings on the Default tab of the input node. |
|
| Back to top |
|
|
| kimbert |
| Posted: Fri Sep 15, 2006 2:12 am Post subject: |
|
|
Jedi Council
Joined: 29 Jul 2003
Posts: 5543
Location: Southampton
|
|
| Back to top |
|
|
| angka |
| Posted: Sun Sep 17, 2006 7:57 am Post subject: |
|
|
Chevalier
Joined: 20 Sep 2005
Posts: 406
|
Hi,
kimbert: My only identifier of each message is in the MQMD Correlation ID. thanks. I will take a look. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Sun Sep 17, 2006 2:41 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20767
Location: LI,NY
|
| angka wrote: |
Hi,
kimbert: My only identifier of each message is in the MQMD Correlation ID. thanks. I will take a look. |
If this is the case and we are not talking about a reply (see MQ Get node in V6) you can use the first flow as routing and keep the correlation Id.
In the subsequent flows use a compute node and use what kimbert told you:
Create ... Domain .... Parse...
This should reduce the number of message sets needed in the single routing node to none ?
Enjoy 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| kimbert |
| Posted: Mon Sep 18, 2006 12:35 am Post subject: |
|
|
Jedi Council
Joined: 29 Jul 2003
Posts: 5543
Location: Southampton
|
| Quote: |
| My only identifier of each message is in the MQMD Correlation ID |
In that case, you cannot use the technique I was suggesting - the MRM Message Identifier needs to be in the body of the message. |
|
| Back to top |
|
|
|
|
