Hi @dylan, my friend @William_B_Schaefer_I tried this out today and he says that the new Roon still doesn’t ever finish analyzing the Qua Continuum Track. Did you guys test that?
Casey
Hi @Casey_Leedom,
We did test this behavior on our end and didn’t have issues with Analysis completing, do keep in mind that a large DSD file may take some time to finish analyzing.
Can you let me know if you are seeing the same thing as William on your end when you have a chance?
– Noris
Hi @Casey_Leedom,
I just wanted to check in here again with you, has the new build of Roon resolved the issue with this DSD track? Please let me know when you have a chance!
– Noris
Hi @Noris, I need to download the problematic Qua Continuum Track again. It’ll probably have to wait for this Sunday. I assume that you guys got a hold of a copy of that track and we’re able to reproduce the issue? Just out of curiosity, can you describe what the bug was? Im a Software Engineer myself …
Casey
Hi @Casey_Leedom,
Yes, we were able to get a copy of the track from Cookie and reproduced the issue. The dev team then looked into the matter and issues a change which has been confirmed as working by our QA dept.
I can’t personally provide a technical answer to this but I have put in a request for the dev team to see what they can comment on regarding the change. Once I hear back, I’ll be sure to let you know.
Do let me know how things are going on your end when you have a chance to take a look!
Thanks,
Noris
Hey Casey – thanks for your patience here!
So, during analysis we generate a waveform that’s shown in the footer. The code that generates that waveform was using a signed 32bit integer to track which sample in the file it was looking at, and because these files were DSD and very long, they contained more than 2^31 samples. That overflowed the 32bit integer, and we tried to write a negative offset into our waveform array.
Hope that helps, and thanks again Casey!
1 Like
Thanks @mike. So did you guys go for uint32_t or all the way to uint64_t “just to be sure”? 
And it is working for me. I no longer have crashes!
Casey
1 Like
Straight to 64bit, because there’s no real reason not to. There’s no good way to get enough confidence that we’ll never see a file bit enough to break uint32,
Yep, these crazy DSD folks are talking about going to DSD1024! [[ Personally, I don’t see any reason to go beyond 96kHz/24bit, but I have … er, “strange” friends …
Casey
This topic was automatically closed 36 hours after the last reply. New replies are no longer allowed.