By "updating Omniscope" I presume you mean clicking "Load" or "Refresh", i.e. taking the latest data from DM and injecting it into Omniscope's views.

Until very recently, the data for all your blocks was only ever held in memory (not counting the source blocks, where the data originates from a CSV file or database, for example). With large models, if you were short on memory, this data would "unload" from memory. The next time a downstream block's data was required, it would need to recreate the entire chain of blocks. Does this fit with what you're seeing?

In the last week or so, we introduced caching to disk of each block's data. This is completely transparent and you should not normally be aware of it. This means that, even if blocks go out of memory, Omniscope does not need to re-execute them, and can retrieve the data from disk.

Those with security concerns regarding caching to disk of their data should be aware that the data is primitively encrypted against casual prying eyes, and caching to disk can be completely disabled by a command-line switch.

So, please try a very recent version of 2.6 from the daily builds link in the download page and see if this problem is resolved.

On a related note, we are working on allowing you to work on a particular operation without triggering refreshes downstream. Instead, the downstream blocks become "stale" and faded, just as with source blocks when the CSV file has been updated, for example. When you've finished tweaking your join operation, you refresh all stale blocks to propagate the changes.

Thanks for this Steve,

I understand what you're saying and have a feeling that this fits to what I am seeing.
I will try this and let you know.

Another related question on this has to do with stopping running processes in Omniscope.
Is there a way we can stop a refresh process or any other proces like resetting, when it takes too long due to limited memory.
In MS Access you have for example ctrl+Break to stop any process started. Is something like this possible in Omniscope, or already there?

Thanks
Arjan

No, you cannot stop a refresh in DM once it has started (although you can cancel a progress dialog if you see one). You must prevent it happening, by stopping auto refresh and operation refresh (as above).

Closing the file or removing the block will, of course, stop whatever DataManager is doing.

From tonight's build of 2.6, see the DataManager view Tools menu, "Propagate block changes downstream" (the location of this setting is subject to change).

When ticked (the default), you get the old behaviour. If you change a block, all blocks downstream recalculate immediately.

When unticked, if you change a block, blocks downstream become "stale". They become faded, with a red refresh button, and do not recalculate.

This should make working with large models far easier.