Re: Using VLC 4.0.0 on WSL2 Ubuntu 25.04 with wayland
Fatih Uzunoğlu <[email protected]> Mon, 23 Jun 2025 16:22:32 +0300
| Newsgroups | gmane.comp.video.videolan.vlc.devel |
|---|---|
| Message-ID | <AM6PR0202MB3368711272B68B41A122FC72AF79A@AM6PR0202MB3368.eurprd02.prod.outlook.com> |
> The SHM module was already using the viewporter in 3.0 The SHM module should handle fractional scale if it is using viewporter. My understanding from the protocol and this crash is that whatever entity using viewporter needs to handle fractional scale as well. This probably means that it should be moved from the interface to the display module to prevent conflicts like this, and certainly the higher priority EGL module should support that as well (not only SHM). > A provider cannot use *any* rendering service on the surface unless it can't conflict with *any* potential use by *any* display. As I said, I'm willing to not use viewporter in the interface, but what is the alternative at the moment? I don't think not supporting fractional scale in 2025 is a good alternative. I used it in the interface because, 1) I'm more familiar with it than the display modules, and 2) the buffer scale was already set there (`wl_surface_set_buffer_scale()`), it only made sense to handle fractional scale there as well. > If triggering a crash in preexisting working merged code is not breaking it, then I don't know what is. We introduced a condition in Qt interface that we use viewporter, like the fractional protocol says, to support fractional scale in order to fit the video window properly. This did not require modifying the SHM module to cause crashes without the interface. This is problematic with the interface embedding in use, and it should be fixed if possible. > No, you just made that up. Can you even point to a precedent for that bold misleading claim? There is no guarantee that window embedding must be possible with all cases. I remember the `EGL_KHR_display_reference` case for example, which was a requirement for embedding. There can be a variety of configuration that can potentially make embedding impossible. This does not mean that these configuration should not exist, just to satisfy embedding in all cases. If in some situations video embedding is not possible, they should not be enabled or offered when the interface is used. I do not know if with the SHM module viewporter is the only issue, but as I said, we should try to support as many configurations as feasible in my opinion. _______________________________________________ vlc-devel mailing list To unsubscribe or modify your subscription options: https://mailman.videolan.org/listinfo/vlc-devel