FYI: Spotify OSS page is out of date about spicetify-canvas

[itsmeow]

It appears the recommended args for automated builds are the following (and likely what Spotify uses as well):

Source: https://bitbucket.org/chromiumembedded/cef/wiki/AutomatedBuildSetup.md

I also found this thread, and I also have been having the make_distrib failure. This is probably it!
https://www.magpcss.org/ceforum/viewtopic.php?f=6&t=18346&start=10

[itsmeow]

README addresses this. You can also just use chrome://version in Chrome Tools.

[itsmeow]

I've had the same problem, I think it's because the branch is on a newer minor version than Spotify. I was trying to build the exact commit but my hard drive decided to crash and I'm giving it some rest. Try using the commit method in the README, let me know.

[itsmeow]

I tried running Spotify on an existing published build of CEF without MP4 (https://cef-builds.spotifycdn.com/index.html#windows32), the version was 90.6.6, Spotify is running 90.6.5. The build worked just fine, so it does mean Spotify is still patchable. Right now 4430's latest is on 90.6.7, which might break Spotify? I will download the 90.6.7 without MP4 build and try it out.

[GeorgeHahn]

Sorry, I completely missed the readme update! I'm running that build now and will report back.

[itsmeow]

It runs on 90.6.7 just fine, so we must be missing some type of build argument that Spotify is not happy about.

I personally thought my issues were just because my hard drive was failing and giving bad data, but maybe my build was fine?
Did you use the new files? They changed the pak files required.

[itsmeow]

Also I noticed, the cefclient on my build doesn't work, so I don't see why Spotify would work with it either. Downloaded builds client works just fine.

Does yours work?

[GeorgeHahn]

I noticed the file change but didn't try to find out which were required - I just dropped my whole build folder into the spotify directory. One moment, I'll test cefclient.

[GeorgeHahn]

Nope, my cefclient is broken identically.

[itsmeow]

Maybe enable_nacl=false blink_symbol_level=0 symbol_level=0 is to blame?

[itsmeow]

I'll add use_thin_lto=false to the README, I think the nacl + symbols are safe for now, since no one explicitly mentioned them. I have no idea what use_thin_lto=false does, but we'll see. Try building that, let me know. I'm not gonna mess with it has my external hard drive with CEF's source on it is on its last legs.

[itsmeow]

Found the related Chromium bug. Weird issue. https://bugs.chromium.org/p/chromium/issues/detail?id=1177001

[GeorgeHahn]

Good find! I restarted my build with that flag set and will keep an eye on the upstream issue.

[itsmeow]

You may be able to use the patch method without a full rebuild. Look at the new section of the README on re-using the CEF build folder.

[GeorgeHahn]

I wish I could; I delete the source directory after each build 😬

[GeorgeHahn]

I'll be back in a few hours, hopefully with good news!

[GeorgeHahn]

Confirmed - a build at 7a604aa without thin LTO works.

I just updated to the latest VS 2019 build tools and will try a build with LTO turned back on.

[GeorgeHahn]

Nope, the same commit with thinlto enabled on the latest 2019 build tools is broken.

[itsmeow]

I don't see why a newer VS would fix it, seeing as the chromium bug is not related to the build tools but ThinLTO itself

[GeorgeHahn]

The build tools include the compiler; an updated compiler may fix thinlto.