<azonenberg>
It's going well, just taking its sweeeeet time
balrog has joined #scopehal
massi has joined #scopehal
<d1b2>
<Darius> that's C++ for you 😦
<azonenberg>
this isn't C++'s fault
<azonenberg>
it's that we have close to 150K lines of code and are making a significant change to a core data structure used across the whole project
<azonenberg>
a change that is complicated enough there's no good way to automate it
<azonenberg>
I knew it was going to be painful. This is the... third, i think? iteration of our internal data model
<azonenberg>
no wait, fourth
<azonenberg>
third refactor
<azonenberg>
But we had more filters and drivers each time
<azonenberg>
so the longer i put this one off the worse it was gonna get
<azonenberg>
the good news is, i think i've finally ironed out all of the kinks and this new data model should last us for quite a while without needing any further changes
<electronic_eel>
let github copilot do the refactor ;)
<azonenberg>
looooool
<azonenberg>
wcgw
<azonenberg>
I'm now up to file names starting with D in scopeprotocols
<d1b2>
<louis> Total garbage displayed trying to render large waveform zoomed out -- about 0.5s x-range here
<d1b2>
<louis> also, FFT filter does not produce any output when applied to this waveform; am I hitting the large-waveform failure case?
<azonenberg>
head *should* work. but there may be bugs
<azonenberg>
fft with >16M points will fail for GPU and fall back to ffts
<azonenberg>
or should, at least
<azonenberg>
you can try --noopencl to explicitly fall back
<azonenberg>
anyway, i hope to finish the big refactoring in the next day or two and then do an extensive test campaign exercising every filter to make sure nothing broke
<d1b2>
<louis> hm, rendering bug is not a refactor issue