Muon Analysis interface no longer loads current run

General
1

In Mantid version 3.8.x the Muon Analysis interface appears to have stopped loading the current run, instead loading seemingly random runs from the past. The is an essential function, and a quick resolution is needed as users are beginning to migrate to alternative analysis programs.

2

Hello,

I’ve had a look at the code that determines which file is loaded when the current run button is clicked and this hasn’t changed since February this year, so the functionality should not have changed since version 3.6.

I’ve just tested it from my desk using musr using a nightly build from the 29th of October, and that loads the correct file.

I’m currently downloading the latest nightly to check that as well, but don’t expect anything different.

Could you give us a little more detail? What instrument are you doing this for? what runs is it loading (and what is the correct answer you would believe?)

Regards,
Nick

3

Hi Nick,

The problems over the weekend occurred on HiFi (but I’ve seen the same issue elsewhere). For instance, I seem to have the same problem just now running Mantid on my own PC reading EMU data … clicking ‘Load Current Run’ loads run 66994; however the current run is actually 68130.

I believe this problem may have been around for some time, but the symptoms have been erratic and have sometimes gone away with a few clicks of ‘Load Current Run’.

Steve

4

I reported this to Tom Perkins / Anders Markvardsen / Raquel Alvarez Banos by email a while ago. Issue 18116 addresses it. It seems that the interface keeps copies of the 5 temporary files, once it loads them. On asking it to load “Current run” it correctly identifies which of the files would be newest and then uses the stale copy if available.

Closing and opening the Muon Analysis interface clears the cache.

So testing needs to be done on a “live” instrument or you need to replace the files in the background while the Muon Interface is open.

5

Thanks James, that’s really helpful especially as Raquel and Anders left for Christmas yesterday.

I’ll take a look at the code to see if I can find out where this caching is happening.

6

@martyngigg and I have found the cache that is responsible for this after your helpful pointer. Martyn is going to change the code so that “Load Current Run” files never enter the cache.