I have been looking at some .NET code a customer has been using as part of a daily batch process and have been testing this with the IQFeed 4.1.1.1 client.

It was my understanding from the IQFeed documentation, that the <CR><LF>!ENDMSG!<CR><LF> message would signify an end-of-message when using the socket interface to retrieve historical data.

However - when attempting to get data for the symbol 'BUYY' - a symbol from the downloadable symbols list, the following message is returned:

!ERROR! !NONE!

!ENDMSG!
!ERROR! Invalid symbol.

!ENDMSG!

This suggests that there can be (in theory) any number of !ENDMSG! strings returned in response to a historical data lookup. The existing code used does not cater for this, and the only work arounds I can see are:

1 - Identify another 'No more data to be received' indicator (API call possibly ?)
2 - Rewrite the .NET socket client class to open and close the socket for every request (Not very efficient use of the socket interface)
3 - Use some form of timer to try and ensure that the buffer will no longer be fed data from the IQFeed client (Not very pretty way around)

The 'error' has not crept up before, so I am unclear as to whether this message format is a new addition.

Regards,
Thomas

When retrieving historical data and you get an '!ERROR! !NONE!' response, your processing state machine can expect one and only additional response set, in this case it happens to be a '!ERROR! Invalid symbol.'. I think the current documentation may discuss this, but may not be too clear. Once you receive that second response, your state machine can reset and request further information through that socket.

Open a manual telnet session and try various commands to test the various responses.

Hope this helps.

Hello Thomas,

I have been able to duplicate this and am looking into the cause. There appears to be two seperate issues here.

For some reason, BUYY is returning an invalid symbol error which is what is making the second issue appear.

This appears to only affect Weekly, Monthly, and Daily History requests that are sent for invalid symbols. This means that it should be fairly easy to predict when it will happen programaticly until this issue is resolved.

Steve,

Does this means that we should not expect !SYNTAX_ERROR!<CR><LF> anymore as an Error from the daily historical socket?

Best,

Jean-Francois

No. This thread is dealing with some requests that are currently returning 2 separate error messages (which is resulting in two !ENDMSG! as well).

If you do not format a request properly, you will still get a !SYNTAX_ERROR!<CR><LF>.
Edited by DTN_Steve_S on Mar 6, 2006 at 08:33 AM

Thanks for the clarification. The documentation does not explain what is associated with each error message type.

I am assuming that we will get the 2 error messages(back to back) for a bad symbol request until a new version is released. Is this a safe assumption?

Is it possible to receive only one set of error message with the current version?

Best,

Jean-Francois

These are fairly Isolated issues and wont affect all cases so most of your requests will return normal errors.

This issue in this thread is brought on by the fact that BUYY has never actually traded and so there is no history data available.

This issue will only arise with symbols who meet the same criteria.

Valid symbols that have never traded.

There is also a similar issue dealing with History requests on this thread:
http://forums.iqfeed.net/index.cfm?page=topic&topicID=922

but once again, it is for very specific requests (based upon time of day and day of the week that the requests are made).

both of these issues will be resolved in the next release.