DSXchange

There's no syntax. DecimalToDecimal() picks up metadata from the output link itself. Of course, this makes it tricky if you want to use this function to produce an intermediate result (that is, as a nested function). In this case, evaluate it in a stage variable that has the appropriate data type (f...

Is runtime column propagation (RCP) enabled?

Preserve Type will work. I believe this is only available in version 8.1 and later.

The best training material extant is the DVD from DSXchange Learning Center.

As far as I am aware that information is not preserved.

I suspect that U is looking at the XMETA database. Regular client tools reflect the integer values in his post - and do not, as he noted, allow for CAST. But these integers report as timestamps in tools such as the Information Server reporting console.

DSXchange is not a substitute for official support or training. We will attempt to answer specific questions if we can, but are chary of such general questions which indicate a lack of training or research.

OK, clearly I did not understand what you had in mind. I understood you had one row in the reference, value xxxyyy.

Your clarified requirement needs two separate lookups.

Neither will work given your requirement. You need a Transformer stage upstream of the lookup stage to combine the two columns appropriately into one. Then perform the lookup using that one.

I don't know that there's a solution but, as a workaround, convert your date to a Julian date (int32 or int64, perhaps) and take the minimum of that.

Oh dear. Life is never that simple. The actual command will differ depending on whether your hashed file is in the "account" (project) or a pathed hashed file, and you may need to perform further setup in the latter case. COUNT is a command from the RetrieVe query language; you may prefer ...

Message handler, as I mentioned.

There is no way to prevent this alert from being generated.

This alert message is one that you can safely ignore (handle with a message handler). What it is telling you is that you are on an SMP system on which Entire partitioning is implemented by putting a copy of the virtual Data Set into shared memory. Because you have specified a partitioning algorithm ...

You need to specify the data type as string if you want leading zeroes and no decimal placeholder.

There isn't one.

In exactly the same way that there is no UNIX command for taking the count of an Oracle table.

A hashed file is how UniVerse database tables are implemented.