azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/glscopeclient/scopehal-apps | Logs: https://libera.irclog.whitequark.org/scopehal
<_whitenotifier-e> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/glscopeclient/scopehal/compare/8f2acf922340...553c6dcdd595
<_whitenotifier-e> [scopehal] azonenberg 553c6dc - DemoOscilloscope: fixed timestamp rounding error
<azonenberg> (gdb) print children.size()
<azonenberg> $6 = 0
<azonenberg> (gdb) print children.empty()
<azonenberg> $7 = false
<azonenberg> begin() also != end()
<azonenberg> so if you foreach() this container it will crash
<azonenberg> because you are trying to dereference an iterator that doesnt actually point to a valid entry
Bird|otherbox has quit [Ping timeout: 248 seconds]
Bird|otherbox has joined #scopehal
* azonenberg debates whether to cry or rip out hair or set gtk on fire
<azonenberg> but hey, it's not segfaulting. finally
<monochroma> *cough*ImGUI*cough* ;)
<_whitenotifier-e> [scopehal-apps] azonenberg pushed 6 commits to master [+0/-0/±14] https://github.com/glscopeclient/scopehal-apps/compare/621158b1c465...517a711e7e35
<_whitenotifier-e> [scopehal-apps] azonenberg ab04a23 - Added "pinned" field to history window. Doesn't actually do anything useful yet. See #283.
<_whitenotifier-e> [scopehal-apps] azonenberg ac83a86 - HistoryWindow: avoid creating duplicate entries and crashing when arming trigger if a waveform other than the most recent is selected in history
<_whitenotifier-e> [scopehal-apps] azonenberg d032992 - Fixed selection of most recent row not working
<_whitenotifier-e> [scopehal-apps] ... and 3 more commits.
<azonenberg> anyway, that's a bunch of refactoring and cleanup of the history window
<azonenberg> fixed a couple of potential crashes, and history is now removeable for individual waveforms rather than "everything prior to time T"
<azonenberg> this will allow deleting of single waveforms from history manually, and pinning of waveforms in history so they don't get deleted
<azonenberg> although neither of those are implemented yet
<azonenberg> Pinning should be done tomorrow. Deletion will take a bit since i have to figure out why my context menu isn't showing up
<azonenberg> (right click messages aren't processed)
<azonenberg> and fixed that. So hopefully i should be able to take care of #283 and #343 tomorrow
<Bird|otherbox> azonenberg: what sort of freakish nightmare of a container type is misbehaving in the way you're observing?
<azonenberg> Bird|otherbox: Gtk::TreeNodeChildren
GyrosGei1r is now known as GyrosGeier
esden_ has joined #scopehal
JSharp_ has joined #scopehal
Fridtjof has joined #scopehal
JSharp has quit [Ping timeout: 250 seconds]
esden has quit [Ping timeout: 250 seconds]
_whitenotifier-e has quit [Ping timeout: 250 seconds]
Fridtjof_ has quit [Ping timeout: 250 seconds]
esden_ is now known as esden
JSharp_ is now known as JSharp
_whitenotifier-c has joined #scopehal
<_whitenotifier-c> [scopehal] azonenberg opened issue #631: PCIe: tweaks to transport layer decode - https://github.com/glscopeclient/scopehal/issues/631
<_whitenotifier-c> [scopehal] azonenberg labeled issue #631: PCIe: tweaks to transport layer decode - https://github.com/glscopeclient/scopehal/issues/631
<azonenberg> monochroma, lain: soooo right now filters in glscopeclient can output one or more time domain waveform streams
<azonenberg> and a single stream of protocol analyzer events
<azonenberg> I'm trying to figure out how to implement a higher level pcie protocol analyzer
<azonenberg> so for example, grouping request and completion
<azonenberg> ideally this would be a decode that *only* has protocol analyzer events
<azonenberg> and no timeline output
<azonenberg> (#366)
<monochroma> basically pcap output?
<lain> tag list (key/value pairs) for protocol analyzer events that allow you to group things?
<azonenberg> well
<azonenberg> it's more, how to fit it into the overall object model
<azonenberg> also thinking about how the UI should work in terms of potentially out of order transactions
<azonenberg> i think what might make sense is to show requests in the order they were issued
<azonenberg> then completions in child nodes of the tree view
<azonenberg> so you expand a request and you can see details of the request packet then the completion(s)
<azonenberg> without regard to when the completion actually came
<azonenberg> but also, how do i make a decode that has no output on the timeline, only protocol analyzer?
<azonenberg> i can make a dummy output you just have to shove somewhere but that's ugly
<azonenberg> if the decode has no outputs at all i think it would end up being garbage collected
<azonenberg> unless the analyzer holds a reference open
<azonenberg> lots of things to think about