It would create a new object.

I want to update already existing object.

Perhaps something along these lines... ClassMethod UpsertXML(xml, wrapper, class, id, Output object) As %Status {set reader = ##class (%XML.Reader). %New () $$$QuitOnError (reader. OpenString (xml)) do reader. Correlate (wrapper, class) do reader. Next (. ...

The %System/%System/RoutineChange audit event is the only thing other than a source control hook that comes to mind. That's handy for detecting source code changes after the fact (when a class is compiled), but not intercepting attempts to change ...

The example in the documentation to %XML.Reader, does not work for you?     #include %occStatus     // Create a new XML Reader class     Set reader = ##class(%XML.Reader).%New()     // Begin processing of the XML input     Set sc=reader.OpenFile(filename) ...

ROLLBACK should revert any changes in data which was done in a transaction, with some exceptions like $increment on Global. You can look at this example. Class User.Test Extends %Persistent {Property Name As %String; Property CalcName As %String ...

Beginner, Caché Objects, XML, ObjectScript I have a XML enabled persistent class and a XML representation of some object of this class (object ID is available). How can I use XML Reader (or some other mechanism) to automatically update this object?   ...

When you going to do it in production, you will stop it anyway for some time.

But just for testing, I think you can extract backup with any empty ensemble on a new server, and use this extracted data on IRIS.

Hi!

Yes, that solution is good. But we need to stop the production to be able to copy the files, and we don't want to to that. We need a way to transfer the data to IRIS without stopping current Ensemble production.

I think it will not be possible to restore backups from Caché/Ensemble to IRIS. I'm sure that in most cases, it is possible to just rename CACHE.DAT to IRIS.DAT, and configure it exactly as it was in Ensemble, should work. And maybe it will work for ...

I think I would run tests and publish the results.

I think this test proves nothing.

Interesting to know underlying reasoning of saying: "Use bitmap index if there are less than X distinct values in common...