With Wayland by default for many systems :-/ I have such mixed opinions about that.
I do believe that Wayland is superior to X in many important ways, especially security and privacy.
However, defaulting to Wayland will cause regressions for many regular users with tools for screen sharing, capture, colour management, etc.
However, we may never solve the Wayland vs X chicken and egg problem if Wayland doesn’t become the default.
Not a trivial problem.
I’m excited about arcan. For me wayland is a step back in freedom, feels a lot like linux and systemd.. something too big and out of reach.
I’ll stay on X until something is way better. What I’d prefer most is a way to wrap X so I can detach from the session (like tmux) and isolate it (for privacy / security). Currently you lose hardware acceleration when you do something like that (afaik).
I recently started using Fedora 33 with Gnome/Wayland. So far I am not a fan of Wayland. Yes X11 needs to be replaced, but Wayland seems to have a much worse, monolithic architecture, instead of the modular architecture that I would expect from a worthy X11 replacement. Combining the display server and the window manager into a monolithic compositor is not great. The plan 9 window system is an example of a modular design. @ilmu mentioned arcan; I will have to take a look at that.
It’s not clear to me that this is intrinsic to Wayland. I think you could right a privilege-separated compositor that provided extensions for an unprivileged WM to draw decorations and manage windows, but didn’t allow it to intercept events or see the contents of other the windows that it managed. The problem in X11 is that the WM can do all of those things and so you get a little bit of fault isolation (though often WMs reparent windows in exciting ways so can’t take over from a crashed WM easily) but no real separation: a malicious / compromised / buggy WM can do anything that a malicious / compromised / buggy X server can.
It’s not intrinsic to Wayland, the Wayland support in Arcan works like that; one to many wayland compositor processes for one to many wayland clients with possibly different exposed sets of registry object types (Wayland being 90ies style asynchronous-OO RMI). With some ‘creative use of game mechanics’, you can have these attach- and reattach- to switch WM models around and survive crashes.
This is both hard and tedious. It also provokes a whole new set of exciting client bugs with the perfect storm that is the extremely flaccid resource management scheme in the protocol itself, coupled with how client implementations converged on the implicit synchronisation and event binning that comes from the cadence of the render-loop of the single-threaded compositor. It’s hundreds of hours of staring at overly verbose traces and pairing with xml-interface descriptions that go into method signatures, but extremely implicit about flow and possible type mutations. I would not recommend it.
Xorg is mostly wired internally for restricting the permitted set of actions of any clients, and mask out event delivery on a per window basis - https://www.x.org/releases/X11R7.5/doc/security/XACE-Spec.html (now find anything as coherent on the matter written for Wayland, I’ll wait..) - including intercepting and replacing returned buffers on window content retrieval requests. The tooling to actually model and enable this is, on the other hand, shall we say lacking. Also, there is a substantial number of clients and user hacks that actually break outright if you’d enable it is much too big of since they were written with the assumptions of most things being permitted (still less breakage than ‘rewrite-in-wayland’ though…).
Thanks, I’ve been meaning to take a closer look at Arcan, the things I’ve skimmed look interesting. I was confused by the GPL comment in the license doc though, what’s the story there?
Can‘t you always just switch back to X from the login screen, when necessary?
Sure. So now you have to know what X and Wayland are, which apps work with which and why, and switch as needed.
This isn’t a forward step for the usability of Linux by the non-technical.
It seems to be the only way to leave the X Windows evolutionary peak, though.
Already seeding! I think this is the 26th release that I’ve jumped on BitTorrent seeding within the first few hours.
I think it was Ubuntu 12.04 when some tracker stats showed two IPs I controlled as being the top seeds for the first 48 hours.
I was not aware that you could run MS SQL server on Linux. There is a guide of how to install on various distros here, https://docs.microsoft.com/en-us/sql/linux/sql-server-linux-setup?view=sql-server-ver15
SQL Server on Linux is a pretty exciting beast. It uses Drawbridge, which is Windows built as a library OS with a thin host layer, so it’s really a Win32 binary running on a really small Windows NT kernel, on top of Linux.
This comes in super handy in dev because we can do things like snapshot a container with a copy of sql server inside, which is shockingly faster and more reliable than using MS’ own tools to backup and restore copies of data and schema.
Are you aware that we could use PowerShell on Linux too? https://docs.microsoft.com/en-us/powershell/scripting/install/installing-powershell-core-on-linux?view=powershell-7.1
I was aware about the PowerShell, I have read it in several places, but I haven’t heard about the SQL Server. I suppose I might have missed it or Microsoft does not advertise it as other products I suppose.