Abhay (guest)
 |
| 06/27/2008 4:21 PM |
 Quote
 Reply
 Alert
|
Hi,
We have written our own decoder for compacted Arca. We had some tough time in decoding messages. The main reason was the messages are not encoded as per the template specified the specs [ I am referring to Version 2.7 , JUNE 18, 2008]. One of the Support guys - David @ Arca informed us that the template is not updated in the way the data is encoded right now.
Please update the specs with the latest decoding mechanism or ask users to refer to the ArcaFast developemet pack [the C code] and not refer the specs for the template in decodng. The template is on page 63-64 in the 2.7 version of spcs.
Thanks,
Abhay
|
|
|
|
|
Brian
Posts:56
|
| 06/30/2008 2:23 PM |
Quote
Reply
Alert
|
Thanks for the inquiry we will get back to you shortly
NYX Data Team |
|
|
|
|
Abhay (guest)
|
| 06/30/2008 3:24 PM |
Quote
Reply
Alert
|
Here are the details of mismatch found during our multicast Arca testing of compacted data during decoding.
For Imbalance message
Here is the sequence in which the data appears in the buffer as against its position in the Presence Bit map.
Relative Presence
Position In Map
Buffer Field Bit
0 Msg Type 0
1 Stock index 1
2 Sequence # 2
3 Source Time 3
4 Volume 5
5 Total Imbalance 4
6 Market Imbalance 13
7 Price 6
8 Scale 7
9 Auction Type 8
10 EXCH ID 9
11 Security Type 10
12 Session ID 12
13 Auction Time 11
This is not as per fast standards. According to FAST, FIELD n must appear in the stream before field n+1. As you can see the position of Volume appears before Total Imbalance in the buffer, Market Imbalance appears before Price and Session Id appears before Auction Time.
Same issue is seen in the encoding of Delete message where the Session ID appears before Firm ID in the buffer.
Relative Presence
Position In Map
Buffer Field Bit
0 Msg Type 0
1 Stock index 1
2 Sequence # 2
3 Source Time 3
4 Order ID 4
5 Buy/Sell Side 8
6 EXCH ID 9
7 Security Type 10
8 Session ID 12
9 Firm ID 11
Thanks,
Abhay
|
|
|
|
|
Nicu (guest)
|
| 07/03/2008 3:00 AM |
Quote
Reply
Alert
|
Hi Abhay,
Why did you write your own decoder? Is it because you were not satisfied by the performance of the "ArcaFast developement pack"? Or because you are using another language than C/C++? Or is the pack buggy?
|
|
|
|
|
joeconron (guest)
|
| 07/07/2008 3:01 PM |
Quote
Reply
Alert
|
The decoder is written in JAVA, but what has that got to do with the fact that the template does not agree with the wire encoding - that is, that the fields are out of order in the encoded stream?
|
|
|
|
|
Abhay (guest)
|
| 07/08/2008 10:32 AM |
Quote
Reply
Alert
|
| We wrote our own decoder since we we are using java. |
|
|
|
|
Nicu (guest)
|
| 07/09/2008 2:48 AM |
Quote
Reply
Alert
|
[quote]Posted By n/a on 07/07/2008 3:01 PM
The decoder is written in JAVA, but what has that got to do with the fact that the template does not agree with the wire encoding - that is, that the fields are out of order in the encoded stream?
[/quote]
Nothing, of course :) I plan to use the C pack and I was just curious if you found any problems with it (and because of that you wrote your own decoder).
|
|
|
|
|