Some Of The Grandest Open-Source / Linux Letdowns Over The Years

[TemplarGR]

opencl 1.2, I mean, am I really asking too much!?

Indeed this is a big letdown. What bothers me is that various vendors in MESA didn't join their efforts with Clover/Gallium Compute and instead decided to just release their own versions out of MESA duplicating effort and wasting years without any proper OpenCL support for the masses. This led to projects not really rushing to use OpenCL much.

At least i hope to see Red Hat improving Gallium Compute for Nouveau and somehow helping AMD users as well in the process. ROCm is good and all but it only supports very recent hardware.

[k1e0x]

Oh there was the Oracle brining ZFS to Linux over the last year. Man Oracle buried that guy. (It's cool I'm sure he found a job working on OpenZFS so this is a community win, not a letdown.)

[Luke_Wolf]

For me the biggest let down so far is that those who make games do not have a really easy way to get Linux users their games.

Sure the major Engines have a "build for Linux" option, but something is holding them back. Either not knowing how to support the platform, or specific problems with porting or choosing to use and customise the DirectX engines for their games or something else.

I know that the devs are getting there and I hope that with Vulkan we have a path for it. It's just that this future is not here yet.

It also doesn't help that the Unreal Engine is a pain in the ass to get up and running on Linux for development purposes and very much isn't a first class experience. It works but you must jump through hoops to get it up and running.

[dkasak]

The biggest letdown for me is GNOME 3.
Giant title bars, air in the UIs, sometimes hard to understand icons (like the "New" icon in gedit), hard to minimize without enabling the icon in tweak tool, you need to jump into Activities to select a different window at times, CSD, extension breakage/slowdowns, etc.

The biggest let-down for me is the trolls.

[torturedutopian]

Oh man, some things never change, do we really have to turn every single thread into a Gnome versus KDE argument ?

I think the biggest let-down is neither KDE nor GNOME but all the useless & time/energy/motivation-wasting splits/forks in DEs. Sure, quite a few of them are due to Gnome 3 but there was a long period of time between the early KDE 4 and late KDE 5 during which it was difficult to advise the regular Joe user to use KDE too. And there was a significant period of time during which Unity was not rock solid either. And the others too. And still today, when using some GPUs, quite a few DEs don't work that well without tweaks. Oh man ! So much because of politics !

I know for devs / geeks / power-users like us, most were usable in most cases... But that was not my point.

[DMJC]

For me the biggest letdowns are (long post coming):

Open source tools for making games: There's almost no asset management tools for Linux, Look at the Descent Manager Tools for Freespace 1/2 and you'll start to understand how stuffed Linux is for this. The lack of anything resembling FRED (Freespace 1/2 mission editor), for space games.

The lack of any real clone of Caligari trueSpace. E.g an easy to use 3D modeller that doesn't rely on the mouse for placing everything (my workflow matters and Blender is not a replacement). I'm now writing my own because no one else would.

The lack of any good 3D model viewers/converters for Linux. E.g where's a small app that loads up my 3D models with textures and lets me spin them in 3D and export to other formats? (think FFMpeg/gnome-mplayer for 3D) Also where is the viewer that lets me select the textures on a model like layers and turn them on/off. Which reloads the textures whenever I make the program window gain focus. One of the best texturing tools is a 3D viewer that reloads textures when the program gains focus, you simply have it and your 2d art program open at the same time on two screens and save/switch from the 2d editor to the 3D viewer to see the textures update in realtime, no complex/buggy integration code needed.

GNOME 3: What the hell were they thinking? Throwing out the 100 million USD of R&D that Sun provided them (I was at Linux.conf.au 2004 where they announced that), the success of MATE, and the absolute failure to gain market share of the GNOME Tablet offerings shows the folly of this move. Noone asked for or wants a unified environment for mobile/desktop. The form factors are completely different. At most people want the ability to launch MATE from their phone when it's connected to a keyboard/mouse. Stop trying to force people into shit UI metaphors and people will use your products.

GTKMM/GTK+/GTKBuilder: Why isn't there an option in Glade to automatically create/bind signal handlers for C++ developers? Why not add the ability to define in Glade if a signal is a member function or not? Glade becomes almost useless for a C++ project if you have to manually plumb up all the signal handlers in C++, a simple checkbox next to the signal handler field would suffice.

WINE: It's been 17 years and I still can't play Warhammer: Dark Omen on Linux, or use 3D Studio Max/trueSpace in a stable manner. WINE has utterly failed at making Windows software run on Linux well. Too much focus is being put on modern 3D games/graphics without actually fixing older software issues. WINE will always be chasing the current DirectX version and never making progress on marketshare/compatibility. Also, WINE doesn't run most corporate software such as Cisco Jabber, any of the Cisco editors, or the VMWare VSphere Console viewers, as someone who has to do real corporate work e.g Phone systems and administration of VMWare clusters, much of it is still only possible on Windows.

[Djhg2000]

I'd switch to KDE of they either implemented GVFS (not gonna happen) or implemented something with similar capabilities (there's like 3 stalled projects trying to do this). I mean I like most of the KIO stuff but GVFS nailed the compatibility aspect.

[Electric-Gecko]

This list is missing something very huge; Longs Peak. This was supposed to be a big overhaul of OpenGL, but the OpenGL 3.0 that ended up being released was far less radical than previously promised. This was because of technical issues with the new object model, which never got released.

Many developers were upset with this, and some even switched to DirectX in protest.

[waxhead]

My list looks a bit different:

- A follow up story about "A decade of wasted cores" - what happened, what got patched? what did not?
- VFS hot data tracking where is it?
- BTRFS lack of 'lazytime' mount option
- BTRFS lack of per subvolume data/metadata profile ("RAID" 1 / 10)
- UEFI : Yes this is a overcomplicated piece of crap, and requires a stupid boot partition. Not nice for us who like to run mdraid / btrfs pooled systems.
- Lack of a large LINUX DEMOSCENE
- Lack of technologies such as Amiga datatypes that allows any program to load any audio/image/video/compressed (XPK libraries anyone?) file in a standard way
- Projects such as LinuxPMI (OpenMOSIX) for clustering seems to have silently died.
- XFCE progress (as long as they don't follow the 'modern' horrible GUI guidelines but stay true to a SANE compact design)
- pfSense was a letdown , talk about assholes. Luckily OPNsense seems to do things in the right direction, but admittedly they are not exactly angels either.
- Minix3, it would have been nice to see the Minix3 kernel with a Debian toolchain. Minix3's kernel have some GREAT concepts!
- Pidgin seems awfully quiet these days...

[Inopia]

I'm disappointed at Canonical for losing their focus on making the best user friendly Desktop+Server Linux distribution and instead pursuing all kinds of pipe dream projects in the mobile and embedded space. Now after neglecting Unity 7 while wasting their efforts on Unity 8 they're just going back to Gnome even though the original reasons for going Unity over Gnome still seem to be valid.

Also when are they going to cut their losses on bzr?

Will snap be a similar disappointment like almost all of Canonical's in-house projects?