waveforms buggy and weird, 1.12 alpha

[RAWRR]

Waveforms don't show at all in mixxx-1.12.0-alpha-master-git4238-x86 in windows XPSP3. On startup, the waveform space is transparent, showing the desktop, which is frozen into it thereafter as the Mixxx window is moved around.

Tried mixxx-1.12.0-alpha-master-git4237-x86 previously and the waveforms were buggy, seemed to "stick" every so often, especially when operating certain widgets. In that alpha the waveforms do actually show, however.

[Max Linke (max-linke)]

Since XP reaches End of Life on April 8 we should consider dropping support for it. We already provide features that require a working internet connection and it looks like there are more to come in the future. Also I don't know if any dev has access to a XP machine.

[RAWRR (rawrr)]

I kind of doubt this is a XP-centric issue, but I can test this on a Win7 machine sometime soon.

However, are we going to also drop support for 1024x768 screens because almost nobody manufactures them anymore? Or reading optical drives?

XP is quite long in the tooth, and it is reaching end of support. But it will remain technologically viable for at least another five years, probably more. I think it at the least deserves a nod and a glance at this point.

[RAWRR (rawrr)]

I'm also curious what non-broadcasting, non-frivolous-id3-data-fetching features won't work without internet, because if there are any, that would be total bs.

[RJ Skerry-Ryan (rryan)]

> Since XP reaches End of Life on April 8 we should consider dropping support for it.

This is temping, but for the past 6 months, 20% of Windows visitors to mixxx.org are on Windows XP.

[Max Linke (max-linke)]

Hey if this also occurs on Win7/8 and I can reproduce it on my machine I'll try to fix it.

RAWRR
Currently only broadcasting and musicbrainz require an internet connection. But we also have bugs to integrate more services (twitter, soundcloud, ...) and GSoC students usually like those and pick them from the bug-tracker or suggest them to us, so it is reasonable to assume that we will get more in the future.

I didn't mean to offend you. We are only a small developer team and do this in our free time. So it is not unreasonable to drop support for a library/OS that is not supported anymore by the original developers so that we have more time to integrate awesome features like the new effects framework.

Since we plan to switch to Qt5 in the next release we likely have to drop XP after 1.12 anyway. Because digia does not actively support XP for it.
http://qt-project.org/doc/qt-5/supported-platforms.html#reference-configurations

[RJ Skerry-Ryan (rryan)]

RE: non-frivolous

automatically downloading or updating MIDI / HID / keyboard presets from a repository instead of having to go to the forums and hunt around

:)

Ugh, why Digia? There will probably still be ways to keep it limping along.

[RAWRR (rawrr)]

Max - Thank you for being gracious and careful, but I wasn't offended. The tone you probably detected in my comment had more to do with my strong feelings about internet reliance issues. I think the "internet of things" is a deeply and profoundly flawed design movement, and I don't feel that software or hardware that is not explicitly made for communication or connectivity should concern itself with the internet any more than my shoe should schedule an appointment with the cobbler all by itself when it senses that it needs to be resoled. I also think webapps are a horrible thing. Most of these concerns are security related.

As long as Mixxx doesn't need to be online to fulfill its primary function, and as long as network connectivity is optional and controllable (including proxy/SSL support, etc.) I'm completely unperturbed.

As for the actual topic of this bug ;) I suppose when Qt makes a decision we'll have no choice but to get saddled with the consequences. I didn't know they'd set that bar. In general, though it seems like good design to be as backwards-compatible as possible. Not just for stragglers or stubborn users, but because working older hardware isn't garbage, it can't all get converted into a webserver or whatever, and honestly I don't think software developers should acquiesce to marginalizing it just because exciting new possibilities exist along the current edge. 20 years of development is embedded in the NT kernel, which is still what Windows 8 uses. I don't think we're going to be in too much trouble supporting XP, even if there is a bit of limping involved:)

[Daniel Schürmann (daschuer)]

Please be focused.

This is definitely high priority Bug for 1.12

WinXP has a market share of 27 %
http://www.netmarketshare.com/operating-system-market-share.aspx?qprid=10&qpcustomd=0

Even If we decide not to support XP one day, we should know what causes this regression.

@RAWRR
* What is your graphic setup?
* Is it possible to identify the commit which introduces the regression
* Which waveform types are effected?
* Are the spinnys effected to0?

[RAWRR (rawrr)]

Sorry. I figured that off-topic stuff was useful to express.

Something weird just happened. I went to test to make sure all of the waveforms failed, as they did for several restarts last night and today, and the waveforms - all of them - were working, though jittery and spastic as I mentioned re: mixxx-1.12.0-alpha-master-git4237-x86.

I did not test the spinnys during the broken waveform period.

The only thing I did between now and the last time they weren't working at all, which was an hour or so ago, was run CPU-Z and dxdiag to get my accurate specs:

graphic setup:

  GPU Intel Mobile Express 965 chipset with 384 MB mem
  DirectX version: 9.0c

latest tested:

  mixxx-1.12.0-alpha-master-git4238-x86

I'm atm trying to relapse the bug.

[RAWRR (rawrr)]

And when they weren't working at all before, it was all of them, without exception.

[RAWRR (rawrr)]

After fooling around with it for a long while I still haven't replicated the original bug, which was "no waveform at all using any type and showing only a transparent window where the forms should be", but I have had other results:

Filtered - QT (GL) is the most buggy. Often one waveform will not show while the other remains. This one also flickers sometimes.
Simple - (GL) is the second most buggy, often flickering.

I got these results by switching between types with the prefs window open, using the scroll wheel on the mouse hovered over the dropdown.

When I was trying to replicate the original bug, I shuffled through all the types in this fashion, and also I switched types and restarted each time.

The easiest bug for me to duplicate is the extreme "glitch/flicker on type change". I actually left it flickering for awhile while trying to set up a screen capture or webcam video of it, and it ran my memory extremely. Mixxx became unresponsive and I had to kill it via Task Manager.

Now Mixxx won't start at all, not even the 1.10 and 1.11 versions that I have installed in other directories.

[RAWRR (rawrr)]

Update: Uninstalled mixxx-1.12.0-alpha-master-git4238-x86 and installed mixxx-1.12.0-alpha-master-git4239-x86. Mixxx will start now. Still has same problems. Current fiddling shows that Simple - Qt (GL) also flickers and glitches. Filtered - Software seems the most stable.

[RAWRR (rawrr)]

Next I will install the oldest alpha and see if it has same problems, but not tonight. Will work forward from there up to alphas already tested.

[RJ Skerry-Ryan (rryan)]

Hey RAWRR,

When Mixxx won't start (presumably because it is crashing) could you do the following steps from this page:
http://mixxx.org/wiki/doku.php/reporting_bugs#windows

1) Enable Windows Error Reporting
2) Collect info from Debug Diagnostic tool
3) Get event logs
4) Do the Windows Error Reporting steps

We're kind of flying blind here so these logs may be the best we can do.

Also, if you can get GDB working on Windows, backtraces would be very helpful:
http://mixxx.org/wiki/doku.php/creating_backtraces#windows

[RAWRR (rawrr)]

I haven't been able to crash it or get it to fail to run again.

As for GDB, it keeps saying "no debugging symbols found".

[Sean M. Pappalardo (pegasus-renegadetech)]

For the record, I still have XP systems, including a 64-bit one, and I'm all about Mixxx running on older hardware. (After a certain point though, that hardware will need to be running Linux. :) )

[RAWRR (rawrr)]

Just tested all of them, (X86, that is) and they all do the glitches and sometimes the waveforms disappear entirely (not just on the empty one), though, again, haven't been able to crash them again yet.

Use the scrollwheel on the mouse while hovered over the dropdown to get this result.

I'm setting up all the debugging crap now, going to try another angle with GDB. Will it still be helpful even if I can't cause a crash?

[Owen Williams (ywwg)]

try trunk again, GLSL got updated

[Max Linke (max-linke)]

Is this still an issue with the current master?

[Max Linke (max-linke)]

Can anyone reproduce this with the current alpha builds?

[RAWRR (rawrr)]

I've been offline for quite awhile owing to viral issues. Don't know when I'll get back into an XP system. Sorry I've been out of this process for so long.

As a side note, I will still need the info asked in my last above question and probably will need some kind of coherent tutorial re: GDB.

[Owen Williams (ywwg)]

If someone can't reproduce this we'll have to mark it incomplete. My own XP machine has never had problems with waveforms.

[RAWRR (rawrr)]

Possibly unrelated, but just for the sake of extra information: I'm having this same series of issues intermittently on my 1.12.0 beta1 5443 Linux install. All of the GL forms show a transparent (i.e. showing the desktop) window where the waveforms should be.

[RAWRR (rawrr)]

(The non-GL forms work fine)

[Swiftb0y (swiftb0y)]

Mixxx now uses GitHub for bug tracking. This bug has been migrated to:
https://github.com/mixxxdj/mixxx/issues/7393