Wireshark-dev: [Wireshark-dev] RTP player redesign in 2.x makes it worse than the legacy one
From: "Peter Budny" <pbudny@xxxxxxxxxx>
Date: Thu, 22 Dec 2016 14:43:34 -0500
Hello, I admit that this is only my opinion, but as someone who uses the RTP Player in Wireshare very regularly in my job, I hope you'll give it some consideration: The UI of the redesigned RTP Player in Wireshark 2.x *sucks*, and its usability is much worse than the old one that's now in Wireshark Legacy. To wit: - The RTP Player (and other RTP windows) permanently take focus. Whichever one on top is the *only* window that's allowed to have focus, and you can't go to another window; you *must* close the window on top. This is baffling, since the RTP Player says you can press "G" to go to the packet under the cursor, but in fact you can't because you have to close the RTP Player and the other windows you used to get there in order to look at a packet! - When playing 2 audio streams, they both play out of the center channel, rather than one from Left and one from Right. Maybe that's better for some users, but it's a lot worse for me. I often need to hear both halves of a conversation simultaneously, in a way that I can hear what audio is coming from which side (e.g., trying to diagnose problems with echo cancellation settings). Playing both streams through the center channel is much worse. - You can no longer choose which streams to play from the RTP Player. There's a table of streams, and you can select one stream in the table to highlight it in the waveform graph (a feature of dubious utility), but regardless of what's selected in that table, it always plays all streams. The only way to play a single stream is to close the Player and the Stream Analysis, and open them back up with only a single stream selected. - Almost everything about the new waveform view is a mess. I can be brief and say "make the graph work like the legacy RTP Player, or like Audacity", but if you want my detailed complaints about it, here we go: - With multiple streams, their waveforms are displayed overlapping each other in the same graph, but each centered around its own 0 y-axis. If they all had the same 0 y-axis, that would make sense mathematically/ scientifically, although I wouldn't like it because one waveform will inevitably obscure another one. If they were on separate graphs, as in the legacy Player, that would make sense and be useful and easy to understand. The weird compromise that the new RTP Player chose is nonsensical, with no basis that I can see in audio engineering (compare Audacity) or any other kind of mathematical graph. - The waveform seems to be scaled to fit 100% height of its y-axis, separately for each stream. That change is *extremely* harmful. When I'm analyzing audio with the RTP Player, it's crucial to know on an absolute scale how strong the signal is. Very often this is exactly the data I need to get: is the signal very soft (mic volume is too low), or moderate, or too strong (mic volume is too high)? Is it louder or softer than the other RTP stream? You can't tell that when there's no y-axis displayed, *and* the y-axis isn't displayed or labeled! - Zooming horizontally (into time) makes sense; that's actually a nice new feature. Zooming vertically (into amplitude) *could* make sense, but not the way it's implemented now. Without any y-axis labels, there's no context for how much you're zoomed. This goes right back to the last point, that scaling the waveform to fit but not displaying a y-axis to indicate the scale is harmful. Moreover, zooming vertically doesn't scale each waveform while keeping the 0 points of their respective y-axes in the same place; it just zooms the entire graph, as though it were a picture rather than a mathematical graph. - The ability to drag the waveform graph left and right is fine, since you can zoom in on it. (It only affects the display, which is non-ideal: you still can't select where to start playback from.) Dragging it up and down makes no sense. Again, it's like someone thought this was a picture and not a graph. It's frustrating that, whatever other technical improvements may have been made in Wireshark 2.x, the redesigned RTP Player is hampering my ability to work, and forcing me to go back to Wireshark Legacy whenever I need to examine RTP data. I'm willing to help address these problems, although I don't know how much I'll be able to help. (I am a programmer, but I do embedded firmware and C, not Qt and C++ and GUIs.) But seriously, I'm willing to work with people, whether it's just giving user feedback and participating in discussions for a re-redesign, or testing changes, or even getting my hands dirty with the code (preferably with some assistance from regular contributors). P.S. I figured wireshark-dev would get the most visibility and discussion going, but please redirect me if I should open a bug report instead or share this somewhere else. Cheers, ~ Peter Budny
- Prev by Date: Re: [Wireshark-dev] dissect_per_constrained_integer() with
- Next by Date: [Wireshark-dev] Extcap limitations?
- Previous by thread: Re: [Wireshark-dev] dissect_per_constrained_integer() with
- Next by thread: [Wireshark-dev] Extcap limitations?
- Index(es):