Pre release version 3.7.0 release candidate 1

[ann0see]

THIS IS A PRE RELEASE!

Softins just released the Release candidate 1 of Jamulus 3.7.0 and we would be happy if you downloaded and tested it for serious bugs (server crashes etc.) before 3.7.0 gets released finally.

Get the latest release candidate here: https://github.com/jamulussoftware/jamulus/releases/tag/r3_7_0rc1

ChangeLog (until now):

• introduced new server lists: Any Genre 3, Choral/Barbershop (Server lists getting full #875),
  also renamed Default to Any Genre 1, Any Genre to Any Genre 2,
  and Classical/Folk/Choral to Classical/Folk.

• central server addresses are now in jamulus.io instead of fischvolk.de (Get rid of fischvolk.de #919)

• show --clientname as first word in title to avoid clipping in Windows task bar (--clientname does not display in Windows task bar #789)

• improved Windows installer, coded by ann0see and doloopuntil (Installer should use public start menu (and desktop) for links  #792, New windows installer #841)

• support http links in the chat dialog, coded by hoffie (Support http links in the chat dialog #879)

• support Github actions for CodeQL, Autobuild, Publish Release,
  coded by pljones, nefarius2001, ann0see, ranfdev.

• more detailed version number of unreleased version, coded by nefarius2001 (Version number of unreleased version #475)

• add new --serverpublicip option to support central servers behind NAT,
  coded by hoffie (Add support for central servers behind NAT #954)

• replaced low-quality background image (fancy skin) of mixerboard, created by passing (Fancy Skin Background is full of compression artifacts #970)

• Escape chat messages at the server. This will no longer allow html formated chat messages due to security reasons,
  coded by atsampson (Escape chat messages at the server #939)

• Highlight the fact that the server is recording in the Mixerboard title (feature-req925 recording indicator #968), coded by dcorson-ticino.com

• Implement new --ctrlmidich syntax allowing to specify fader, pan, mute,
  solo buttons (Feature midi pan issue 95 #945). Implementation for mute&solo buttons for now is
  only for toggle controllers and does not support headless operation yet, coded by dakhubgit

• Move ASIO Setup button on Windows below driver selection (Move driver setup button under driver selection #976 #977), coded by mulyaj and ann0see

• Gray out input levels and show a message if you're not connected to a server, coded by ann0see (When not connected: "Please connect to a server to hear yourself and others" #847, Add connect to server help message #983)

• bug fix: remove ConsoleWriterFactory - to resolve suspected memory leak, coded by pljones (#923 Remove ConsoleWriterFactory #926)

• bug fix: fix crash if settings are changed in ASIO4ALL during a connection (contained in Loud noise if Input is disabled in ASIO4ALL while still selected in Jamulus #796). Reverts MacOS: If an audio device is no longer available, show an error rather than switching to Default #727 for Windows

• Avoid screen from sleeping or starting screen saver for Mac, coded by jerogee (Turn off screensaver if connected to a server (LXQt) #834)

• stop sending traffic to CloudFlare as part of the external IP detection
  logic (Use a different approach to get external IP address #633, Server: Stop opening a TCP connection for external IP detection #1092), suggested by atsampson, coded by hoffie

• bug fix: properly restore mixer levels after reconnecting to a restarted
  server. This is a server-side fix. coded by hoffie (Restarting the server on the fly ignores set levels and mixes everybody at 100% #955, Server: Send empty channel list on connect #1010)

• Improve update version detection, coded by softins (Improve version update detection. #1155)
  Check two servers instead of one (updatecheck1.jamulus.io and updatecheck2.jamulus.io).
  Ignore the server version if it is not a release, due to a suffix such as dev, alpha or beta.

• Windows installer interface now supports English, Dutch, French, German, Italian, Polish, Portuguese, Spanish and Swedish.

• bug fix: Fixed some memory leaks in server recording, coded by softins (Fix memory leaks in Jam Recorder #1073)

• bug fix: Adjusted some accelerator keys in translations to avoid duplication (Avoid conflicting accelerator keys. #1165)

• Release now builds .deb installation packages for Ubuntu/Debian, coded by ann0see (Build deb files on Ubuntu 18.04 #1100)
  Compatible with Ubuntu 18.04 or later.

• Release now builds experimental .apk installation packages for Android, coded by nefarius2001 (Automated build for Android APK (updated) #880)

• bug fix: Added missing driverInfo initialisation in Windows sound driver, coded by npostavs (windows/sound.cpp: Explicitly set driverInfo to 0 #1194)

[hoffie]

Linux client has to be investigated further. There might be glitches if you switch between normal and fancy mode. Windows is a bit slow (but this could also be my impression only).

So are you suspecting some regression? Been using the master code (~beta3) both on my servers and on my Linux clients for some time now.

[ann0see]

Probably. But it might also be a glitch based on something odd with my PC or GH Actions.

[ann0see]

I'll compile it on my linux machine now and check if the same bug comes up there.

[ann0see]

Ok. See: #1217 (comment)

It seems to be a regression

[hoffie]

Saw some English "Yes" on buttons for the German installer.

Thanks for reporting.

I can verify this when running the Windows installer under Linux in wine with LC_ALL=C and selecting German in the UI (it's the ASIO-not-found popup).
I cannot verify this when running the Windows installer under Linux in wine with LC_ALL=de_DE and selecting German in the UI.

I suspect that these strings come from NSI-native translations (or even from the OS?) as they are not in our language files.

Can you share details about your environment? Were you trying on Windows or with wine (like me)? What Windows version? Is your Windows language German or English?

[ann0see]

These buttons should be translated automatically. I doubt, we can do anything here.

[wschell]

I have made a public testserver with 3.7.0rc1 in Düsseldorf, Genre Rock, named Testserver 3.7.0rc1.

[ann0see]

@wschell

I have made a public testserver with 3.7.0rc1 in Düsseldorf, Genre Rock, named Testserver 3.7.0rc1.

Which OS are you running it on? Headless Linux/Windows?

[ann0see]

Great. Thank you for the information. So you used the official .deb file?

[wschell]

No, sorry - I made a "git pull" and make...
But I could install the package if this is useful for you. It would be later in the evening.

[ann0see]

Ok. I'd prefer the .deb. Note that it installs to a different location.

[wschell]

Now the headless.deb package is installed and both server are running and registered in genre Rock;

[ann0see]

Great! Thank you very much! I hope everything works fine with them.

[softins]

Maybe I mis-remembered:

Now I could reproduce it like that:

1. Start Jamulus with fancy mode enabled

2. Switch to normal mode while settings is open

3. Switch back to fancy mode (maybe you'll need to switch back and forth between the different modes).

I just set up a Debian 10 VM under VirtualBox on Windows. I installed the .deb from the 3.7.0rc1 assets, and set up jackd on a dummy device (unfortunately, the real device was fixed at 44100).

I could reproduce the issue with exactly those steps.

Once the white panel was showing, I could switch to normal or compact mode ok, but switching back to fancy continued to give the white area.

But while the white area was displayed, if I connected to a server, the fancy skin was restored properly as soon as it showed the fader(s). This was true whether the server had recording enabled or not.

Once that had happened, I could disconnect, and then still switch between Normal and Fancy without any problems or white box.

[softins]

So, I complile 3.7.0rc1 and it shows the bug repeatably. I compile 3.6.2 and it doesn't. Now using git bisect to home in on where it broke.

[ann0see]

@softins we've already found the commit. Also see the linked issue.

Good to see that I'm not the only one with this issue

[softins]

Ah, ok. I'll corroborate then! ;)

[dcorson-ticino-com]

@softins Thanks for the additional information. We are trying to zero in on the problem here: #1238 (comment)
Can you confirm that there are other Linux versions where this is working OK?

[jujudusud]

As far as I am able to check, this RC1 is ok on Archlinux with LTS kernel.

[ann0see]

Great thank you!

[pljones]

Don't know whether there's any point addressing this before release. It might be nice to move "[ ] Enable Small Network Buffers" over above the network readings ("Audio Stream Rate", etc) or to the top of Misc.

[softins]

Actually, I don’t understand why we need the checkbox at all. As far as I have been able to determine, it has no effect when buffers is set to 128 or 256, and if it is not checked, 64 actually behaves the same as 128. So really, it should be implicitly on for 64, and off for 128 or 256. Have I missed anything?

[dcorson-ticino-com]

I don't know except for the fact that there are (or maybe were in the mean time) older servers that did not work with small buffer enabled and 64sample, but did work with 64sample and small buffers not enabled. Because of that I thought the positioning is good.

[pljones]

It's needed because it causes problems sometimes. If you're playing bass, for example, you might need to turn it off. If you switch over to lead guitar, you could probably benefit from turning it on. (Yes, it's noticeable, reproducible and Volker was aware of the point when it was introduced. It's an artefact of passing low pitch tones through the OPUS codec - more cutting causes more "sudden steps" to be possible.)

[softins]

What I mean is that it seems turning off SNB and remaining on 64 is identical to just switching to 128.

[pljones]

Mmm... I guess so -- 64 LPCM samples arriving and being demanded at twice the rate as 128 OPUS samples departing and arriving.

I think the point is that any other audio buffering in the system is at 64 samples, even if the OPUS/network buffer is 128 samples, if that's your audio buffer size, potentially keeping your overall latency down.

Possibly it could stay in "Soundcard" but move under the buffer size and only be enabled if the buffer size is 64 samples? (Note, if I'm using ReaRoute with Reaper set to 64 samples, changing buffer size in Jamulus is disabled.)

Anyway, if it needs this much discussion, it's not for 3.7.0.

[ann0see]

Locked because this is outdated