New user, odd problems

[kimjohn1]

In the past 4 days I have installed SCIRun 5 on several computers operating with Ubuntu and Linux Mint (various versions). All are working fine, so congratulations on whatever changes you have done recently.

I am working my way through the Basic Tutorial, and have discovered that attempting to save a network using "save as..." causes the application to become unresponsive. About half the time the computer system becomes unresponsive, and requires a hard crash recovery (power button held in the off position until the system powers off). Can you think of anything that might be causing this, or what I might do to investigate or work around the issue?

Also, I notice that execution takes 10 to 30 seconds to update the view after making a change in the network. Is there a way to tell the application to use all available cores during operation, or is it already doing that? Are there operator techniques I can use to speed up operation? I am working on getting a more functional GPU for the PC that will be used for SCIRun investigation. Do you have recommendation for what to get?

Kim

[dcwhite]

First question, is this on a native Linux install or one of these emulator setups you mentioned? That might severely impact disk speed for network saving. I have noticed some performance problems on Linux recently though, in some very basic places like latvol computation and ShowField. Those are new since I fixed the build. Your GPU weakness is probably contributing (how much video RAM do you have?), but since these issues I've seen impact a lot of basic areas I'm definitely going to investigate soon. As for the network saving problem, I have not tested that, and seems more likely due to the setup, but it could also be a new bug. I will take a look this week and see what I can replicate.

[kimjohn1]

Correction: the network I loaded for the verbose capture was the Basic Tutorial, Chapter 2, engine.nhdr exercise.
-KJ

[dcwhite]

#2244 should be merged into master by the end of today. If you want to test it right this instant, you can git checkout disable-stop-code

The file load at the start of the file looks pretty fast. That ShowField execution ought to go back to normal with #2244 fix.

[kimjohn1]

All good! My first opportunity to give it a whirl will be this evening. I'll report what luck then.
Kim

[kimjohn1]

Success here with the SCIRun-master retrieved from github at 9:30 pm, 19 March. Operation is smooth and quick. I still have about a 10 second delay when I open ShowField Options and click on Show Edges, but I presume that is some specific issue not addressed by your latest fix. Selections made with ShowField Options for Show Faces and Show Nodes are instantaneously rendered in the ViewScene display. All other view affecting changes that I have tried have been instantly shown in ViewScene when they are applied. Very nice.

I have experienced a bit of strangeness with the ShowField option for Enable Transparency with Edges and Faces. The Enable Transparency option is selectable, and has an effect, but the amount of transparency is not affected by the enterable numeric value. The feature appears to be using some constant value instead of the entered value. The Enable Transparency function for Nodes does not suffer this behavior and appears to be fully functional.

So, this issue is resolved as far as I'm concerned, and I'm pressing on with using SCIRun 5 as the topic of a term project.

Kim

[dcwhite]

I think there's a bug for that--I'll try to find it in the list.