This problem happens all the time and always comes up when Im trying to show someone how rad VVVV is. Something I’ve noticed is that 99% of people don’t have the capacity to see past even an aesthetic flaw and this means that they see this and even at a subconscious level they have already rejected the entire platform. I feel blessed I’m in the other 1%, but its a shame to lose all that art and potential $$$ because of something so silly. Agree with @catweasel above, something like that would be awesome.
It should be a flag, with refresh enabled by default.
If you’re prototyping, then ok, performance hog or not, it doesn’t matter too much. You’d save time a lot of time anyway. You’d just have another vvvv instance (with refresh disabled) to test real performance, if you really must.
Else (when you don’t need GUI), you’d just add the flag with the other ones.
If you’re live patching, well, you’re still stuck, so to say.
latest alphas have UI redraw on F5.
I guess that helps those people that have already gotten used to it, to just more quickly deal with this …
It won’t help with what Chris and grey.phoenix pointed out, though.
F5 will also “break” some patches. I know a lot of people/agencies that use F5 as standard for reloading external settings and/or assets.
I’m thinking i’m also using F5 on some patches crap ;]
Ctrl something will be better
would you say this is an observation you made that it often happens actually when you scroll?
@bjoern @antokhio so are you suggesting we should vote for the shortcut? F5 is just the quickest and most obvious i could think of since it conforms with how browsers do a refresh. don’t you think that “breaking” change could be easily “fixed” by those affected?
@joreg Well, its as much as scrolling past a name thats broken will sometimes fix it, othertimes it will leave a long timeslice trail. Scroll as the refresh would work for me, as it doesnt require a shortcut, and its what you do to fix and sometimes cause the issue. Also I use f keys all the time to switch patches, effects, change focus, they’re otherwise useless keys so are the first ones I use to do stuff with.
@catweasel so you’re saying get rid of F5 and only do it optional on scroll-end?
Well it would work for me, and the others using f keys, but I’m not saying its the only or best solution!
Just another observation: The bug occurs to me especially in large projects and part of the problem is definitely having "CTRl+T"ed some windows. For example if I have multiple windows in as tabs and one is or was brought to the foreground with “CTRL+T” it is all but garuanteed to trigger the refresh bug.
@mburg, i never use Control + T and can provoke the bug easily on a Surface Book with two-finger trackpad scrolling. Scrolling seems to cause the bug, less when using a mouse scrollwheel, more often when using a trackpad with gestures.
anyway, refresh on scroll end sounds good. let’s see if this helps…
ouright, so let’s see where that refresh-on-scroll-end gets us. it is now in place in latest alphas. it adds another spike in the perfmeter, obviously, and since i couldn’t reproduce the problem it tries to fix i cannot even tell if it makes a difference. so we’ll have to wait for your reports…
in the meantime if you don’t like the spike, you can opt out via commandline: /norefreshonscroll
and you can opt in to refresh on F5 via: /refreshonf5
This is the best news I’ve heard all year! I will give the alpha a workout on a patch that constantly shows the problem ASAP. Thanks @joreg!
In my current patch the bug occurs when Alt+Tabbing out of vvvv … the moment the windows alt+tab dialog is shown, I can already see the labels gone missing in the background.
This works in 1 out of 3 times I guess, and somehow happens more likely if I open a subpatch, hide it, and then try to trigger the problem with the containing patch in focus (that now also has a ~ flag after showing said subpatch).
Seems like this is part of the culprit, will check alpha though.
@readme that is exactly the case i was able to reproduce. sometimes every 3rd time, sometimes after 100times. random. anyway that is when i installed the RefreshOnActivate/Deactivate. so you’ll still see the error but clicking back to vvvv should then solve it.
also weird things happen if you scroll in a (large) patch who’s window is partly outside of the desktop. or is that a different bug?
can you show us that? screenshot/recording?
It’s hard to reproduce (as usual), but I will make a screenshot when I come accross it next time. What I mean is when you i.e. scroll down and instead of seeing the next 50 (or so) pixels of the patch, it just keeps repeating the bottom 50px. Maybe thats what @catweasel was referring to as ‘timeslice trail’ (or slit-scan). This can often be fixed by moving the patch window inside the borders of the windows desktop.
@veevee ah right, that is unrelated to the IOBox refresh discussed here.