Closed zsperry closed 5 years ago
Sorry about that Zach, I changed the internal representation of spike waveforms from int16 to single precision format. The change improved the sorting accuracy while increasing the memory use by 25%. I didn’t check for the backward compatibility. I will make it backward compatible.
-James
On Jul 7, 2019, at 12:11 PM, zsperry notifications@github.com wrote:
I know you said you were letting go of some backward compatibility stuff, but just fyi my older prm files now throw an error at nanmean_int16. If there's a way to restore compatibility, that would be awesome, but otherwise I'll just continue working from the 6/17/19 version that still works and still has the features I need (driftview in particular).
Error details:
Index exceeds matrix dimensions.
Error in irc>nanmeanint16 (line 10202) trWav(:, viSite1a, viSpk) = tnWav0(:,viSite1b,viSpk); % P.miSites(:,iSite_unique)
Error in irc>cluwav (line 10146) mrWav_lo_clu1 = nanmeanint16(tnWav_(:,:,viSpkclu), 3, fUseCenterSpk, iSite_clu1, viSiteclu, S0.P);
Error in irc>S_cluwav (line 10048) [mrWav_clu1, viSite_clu1, mrWav_lo_clu1, mrWav_hi_clu1] = cluwav(Sclu, tnWav, iClu, S0);
Error in irc>S_clu_updatewav (line 4586) S_clu = S_cluwav(S_clu, [], [], S0);
Error in irc>manual_ (line 5586) S0.S_clu = S_clu_updatewav(S0.S_clu, P);
Error in irc (line 277) manual_(P); — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.
resolved in v4.9.1
Hmm...still a problem for me...
Sorry, let me find one of the older files you shared with me and test on that.
-James
On Mon, Jul 8, 2019 at 8:29 PM zsperry notifications@github.com wrote:
Hmm...still a problem for me...
— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/jamesjun/ironclust/issues/20?email_source=notifications&email_token=ACEHBOBMHQBDEDTYRMIC3CDP6PLXZA5CNFSM4H6WBCHKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZOXJCA#issuecomment-509441160, or mute the thread https://github.com/notifications/unsubscribe-auth/ACEHBOBMVBHDSQPO2K46Q4LP6PLXZANCNFSM4H6WBCHA .
I know you said you were letting go of some backward compatibility stuff, but just fyi my older prm files now throw an error at nanmean_int16. If there's a way to restore compatibility, that would be awesome, but otherwise I'll just continue working from the 6/17/19 version that still works and still has the features I need (driftview in particular).
Error details: