Desktop Linux Needs a Year of Boring Fixes

LinuxLinks gives me space to share my opinions about Linux, open source and whatever else catches my attention. The views and prejudices are my own.

KDE Frameworks 6.32 caught my eye because of a fix to the way Dolphin copies files to USB storage. The release, which landed on 13 November, addresses a problem first reported in 2011. Dolphin could announce that a copy had finished while the USB stick was still writing the data. Fifteen years is a long time to spend arguing with a progress bar.

I’d like this to set the tone for desktop Linux in 2027. Put the optional interface redesigns on hold for a release cycle and give the awkward, familiar problems some proper attention. I’d happily use a desktop that looked much the same as last year’s if it meant fewer interruptions to my work.

What should desktop Linux prioritise in 2027?

View Results

Loading ...

Finished ought to mean finished

KDE developer Méven Car has written a detailed account of the fix. Dolphin was reporting progress as data went into memory. The kernel still had to write that data to the device, and on a slow USB stick the difference could be considerable. In one of Car’s examples, the stick was still writing for more than a minute after Dolphin considered the copy finished.

You should still eject a drive properly before unplugging it. But the copy notification ought to tell you what’s happening too. Nobody should need to understand Linux’s page cache to make sense of a message saying their files have been copied.

Dolphin now waits for the final writeback to finish before reporting success. That matters even more when you move files to removable storage. The originals stay where they are until that final write succeeds. That’s a detail I care about when my documents are involved.

Car also explains why getting this right took some persistence. The obvious approaches could make transfers much slower, and a solution that worked well for one USB stick also had to cope with other devices and network filesystems.

Give the familiar problems a turn

Think about the things that make you hesitate before recommending Linux to someone. Will their laptop wake properly? Will their monitors come back in the right arrangement when they plug the dock in again? Will they be able to open a shared folder without spending an evening investigating it? Those are the things I’d want a desktop project to put near the top of its list.

A workaround can get you through today. Having to remember it six months later is another matter. Once you’ve learned which service to restart or which setting to toggle, it’s easy to forget how unreasonable the whole arrangement looks to somebody encountering it for the first time.

I have very little patience left for treating that knowledge as part of the normal cost of using Linux. I enjoy knowing how my computer works. I also want to get on with things. There’s no contradiction there.

I’d judge the next desktop release by following ordinary jobs through from beginning to end. Copy a folder containing thousands of small files to an old USB stick. Browse a network share while a large transfer is running. Put a docked laptop to sleep, disconnect it and wake it up at the kitchen table. Can you carry on where you left off?

Do the same jobs using only the keyboard, then try them with a screen reader. Being able to open a dialog is precious little use if you can’t reach the button you need or tell what it does. These checks belong in routine testing, alongside the configurations that happen to work well on a developer’s desk.

Make a dull release worth celebrating

Saying “fix the bugs” is easy, so here’s what I’d actually like to see. Pick a short list of problems that cause people to lose work or repeatedly interrupt an ordinary task. Publish it. Explain where the difficulties are. Give the people working on those problems time to finish, test and review the fixes.

The age of a report is a poor guide to its priority. Some old reports affect very few people. I’d sooner know how often a fault crops up and what it costs the person who hits it. A familiar fault that regularly disrupts a common task deserves attention even if experienced users have become good at avoiding it.

Volunteers will choose work that interests them, and a designer can’t simply be handed a storage bug to fix. The people allocating paid development time, grants and testing resources do have choices, though. I’d like reliability to get first refusal on those resources.

Distributions have a part to play here. If I reported the problem on my distro, the update I care about is the one I can install. Once the fix reaches the distro, try the original reporter’s steps again. Does Dolphin now say the copy is finished at the right moment?

I’d like to give more of these fixes proper coverage on LinuxLinks. Explain what used to go wrong and show what happens now. Credit the people who did the digging. A reliable file operation might be harder to illustrate than a new desktop layout, but a USB stick still writing a minute after the copy supposedly finished is a story I can tell.

There will always be worthwhile new features to build. Some make the desktop usable for people it couldn’t previously serve. Those deserve a hearing. Another cosmetic reshuffle should face a much tougher test when familiar tasks still need work.

If the next release looks much like the current one and saves me a few wasted evenings, I’ll consider that a good upgrade. Give me a year’s worth of those.

What would you fix?

Which everyday Linux problem would you put at the top of the list, and what would a proper fix look like? Tell me in the comments.


All my articles
Linux’s Biggest Problem Isn’t Microsoft. It’s Linux Users.
Distro Hopping Is a Hobby. Stop Pretending It’s Productivity.
Linux Deserves Better Than Your Oldest Computer
Linux Has Too Many Half-Finished Apps
GNOME Has Forgotten That Desktop Users Want Choices
10 Useful Open Source Linux Applications I Miss
AI Companies Owe Open Source More Than a Thank You
I Use Linux. I’m Tired of Making Excuses for It.
Linux Will Never Succeed on the Desktop Until It Stops Expecting Users to Be Experts
Wayland Has Won, But It Hasn’t Replaced Everything X11 Did
Linux Has Too Many Distributions – And Most New Ones Don’t Need to Exist
“Linux Supported” Is One of the Most Misleading Claims in PC Hardware
Why A 50 TOPS NPU Is Almost Completely Useless
If AI Wrote Your Code, Just Say So
Why I moved over to CachyOS from Manjaro
Why desktop Linux should prioritise everyday reliability and bug fixes over new features in 2027.
Subscribe

Please read our Comment Policy before commenting.

Notify of
guest
0 Comments
Oldest
Newest Most Voted