[Laura]

The behaviour of q primitives and KX namespace objects occasionally changes. Most such changes are extensions: what was formerly an error now does something useful. The Reference does not always note in which version the extension was made. If your version of q does not behave as described, check the Release Notes.

Very occasionally a change breaks earlier behavior. The Reference documents such changes; generally for one major version. Currently, the Reference should show any breaking changes introduced since V3.0; for anything earlier, see the Release Notes.

[Laura]

Hey David, thanks for the help!

I’m dealing in degrees, or trying to.

Does kdb sin/cos only accept radians?

For example, if I go sin(30) on my comp calc i get .5. When I go sin(30) in kdb I get -0.9880316.

At it’s simplest form, I’m trying to use the sin/cos/tan functions within KDB similar to how i’d use them on my graphing calculator.

[Laura]

Hi David,
To elaborate on @ekallivrousis comment, we are trying to use 128-bit floating point data type (instead of 64-bit(8 byte double) ) to support 24 decimal places instead of 16 decimals supported by double. Currently, as per C API Reference KDB+ supports 4byte(float) and 8byte(double).  Is there were way to store 128-bit floating point values ?

[Laura]

Thank you David! This is definitely what I was looking for!!Great!!

Richard

[Laura]

Hi David,

Thanks for your kindly reply. I want to use functional query, because I have in each table a field(time-date) to be converted into timestamp. So, instead of using something like this:

update "p"$datefield from dictionary[`table1] 
update "p"$datefield2 from dictionary[`table2] 
update "p"$datefield3 from dictionary[`table3]

I wanted to use this:

coldic:`table1`table2`table2!`datafield`datafield2`datafield3 {![`dictionary[x];();ob;(enlist `coldic[x])!enlist($;"p";`coldic[x])] x} each key coldic

But it is not updating the tables, this is the output :

“” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” “” ..

I’m sorry for not being clear in my concern,

Thanks,

Richard

[Laura]

Hello again
It turns out that

1) reinstalling KX solved the problem and

2) we can run Control as HTTP only.

Unfortunately, though, I could pinpoint what was causing the problem to begin with. Anyways, I will just be happy that the problem is gone!
Kind regards,
Adriano

[Laura]

Hi MSHK, thanks for your question!

Have you seen the documentation on streaming queries at: https://code.kx.com/dashboards/datasources/ ?

Further down the page there is more information on Subscriptions using the streaming or auto-execute functionality.

Hope this helps!

Laura

[Laura]

Hi Adriano,

I am fairly sure you can have TLS kdb enabled and run tomcat in HTTP, although I haven’t tried it myself.

Let us know if it works! If not, we’ll get you to run the support scripts: https://code.kx.com/platform/deployment/raising_support_issues/
and will reach out to the support team for you!

Thanks,

Laura

[Laura]

Hi Leah,

 

Thanks for the quick reply. We will try to redeploy today. However, I have a question:

We install KX with Control on HTTP only mode, but other processes still can run on TLS if required. We were wondering if this might a be a problem as well?

Cheers,
Adriano

[Laura]

Hi Leah,

Thanks for your kindly reply. So, I’m trying to do this in a window environment btw, after making sure that my openssl files are in Path. Tried to run  q -u 1 -E 1 -p 5000 but it didn’t work.

 

For windows, is something else missing?

Thanks,

Richard

[Laura]

Hi Leah,
Thanks for getting back to me and let me know how the export order is determined Hopefully this will be addressed in the future.

For whoever is reading and is interested: my interim fix for this is: I just reorder the exported file by hand using git diff on VSCode whenever I see functions that were reordered automatically by the exporter.

Cheers,
Adriano

[Laura]

There is another ODBC connector available that you can try if the Simba one doesn’t work:

https://code.kx.com/q/interfaces/q-server-for-odbc3/

[Laura]

Update: I did more investigation and found out .z.exit and that solved my problem.
Inside the initialStateFunction of the RTE, I assigned .z.exit to a function that calls the stopServiceCSV.sh script, thus eliminating the need of  using timers

[Laura]

Hi David,
Thanks for the reply. It looks like what I already do is in line with what you told me.

Here is my scenario:
I have one workflow that contains a single RTE that I use to start a bunch of service classes.
To do that, i used the initial state function of the RTE to call the startServiceCSV.sh script provided on the KX Data Warehouse tutorial.  Meanwhile, the csv contains hardcoded information about the “parent” RTE at each service class row.

Then, each service class has a custom initStateFunction that, among other things, creates an IPC handle to their parent RTE and keeps checking (with a timer) if the handle is still valid. So whenever i stop the workflow, the RTE obviously dies, making the handles invalid and therefore killing the services.

I asked about a stopStateFunction because if such a thing existed, i would not have to do the ipc logic – but what I am getting from your response is that this is as good as it gets?

cheers,
Adriano

 

[Laura]

Hi Adriano!

This is an issue not only in IHDBs but RDBs i.e. if EOD is missed due to downtime then not only is the old data never pushed to HDB but it’s no-longer loaded to the RDB (because a new log file is made).

As a custom solution to this issue you could

• check for this scenario on startup
• playback the previous log file (I’m not very familiar with IHDBs but I imagine there’s an equivalent) https://code.kx.com/q/kb/replay-log/
• EOD with the appropriate date

Obviously this falls apart when the process is off for more than a day (you could loop through each missed log file I suppose), but hope that helps!

Josh