10 ms·

A kernel update broke my stylus

213 points by yorwba 3y ago

thanatos519 3y ago
Sounds like someone broke userland!
- Aleklart 3y ago
  “No, we fixed the glitch.”
  - tux3 3y ago
    That doesn't normally fly in kernel land If a 'fix' broke something (and it's not for something like a major security issue), then it is customary to revert the fix and go back to the drawing board I can see that this is a complicated situation and OP was a special case that had not been affected by acccident, but those are excuses If you upgrade your kernel and you don't have any out of tree stuff, by policy nothing should break, orherwise it's the kernel's problem.
    - smelendez 3y ago
      > go back to the drawing board I see what you did there
    - appleskeptic 3y ago
      “Never break userland” is more of a strong guideline than a hard rule, even if Linus has sometimes made it sound otherwise during a heated gamer moment. The fact is, kernel devs break userland code all the time, sometimes because the alternative is worse, or sometimes because a bugfix broke something and no one noticed until much later and it was decided not to revert the fix because it would now break other things to do that. Try to run an entire large userland from 2001 on a modern kernel and tell me how that goes.
      - tux3 3y ago
        Mistakes are human. Certainly a lot of bugs slip through, and that's understandable, software is hard. That being said, I find that people tend to take the no regression rule pretty seriously, if you bring a valid complaint. It makes good sense. We don't want to give people a reason to stay on old kernels even more than they do today (this post written on an ancient Android kernel, chock full of out of tree drivers). We can at least save people from pain when it comes to in-tree drivers. I think we really all benefit from that policy. Even if it sometimes keeps some less than perfect workarounds in the code
      - dfc 3y ago
        I always assumed it was a rule. I don't know of any kernel regressions that broke something I used daily. Are there any examples you can think of?
    - jeroenhd 3y ago
      I mean, for whatever stupid and obscure reason, the manufacturer decided to hard-code the eraser button like that. The device was designed to have two erasers. If you used that tablet on Windows and switched to Linux, you'd notice the device wasn't working like it was supposed to. I can see why one would consider that a bug. One user asks "why is my right click an eraser now", another asks "why is my eraser a right click on Linux". Which one is right? To resolve this, the devs could've added an easy setting somewhere to control this behaviour, but I still think this can be considered a reasonable bugfix as the manufacturer decided on this stupid design. Linux drivers used to hack around that, but that's a classic XKCD 1172 problem.
- tedunangst 3y ago
  Every userland program is equal, but some are more equal than others.
kaetemi 3y ago
Now Windows isn't alone in changing or resetting drawing tablet settings on random updates.
- DarkmSparks 3y ago
  difference with Linux is this will be fixed most likely before new year. windows will always stay broken.
  - m463 3y ago
    windows may be fixed (but the update will also fix your default browser to be edge)
  - pjmlp 3y ago
    Depends, my wlan driver on Ubuntu still has connection issues with my router, it is the only device in the house that requires being occasionally plugged into LAN cable. It is also the only one using GNU/Linux, the TV/WebOS and Android/Linux devices, alongside Windows, just work.
pizzapim 3y ago
A smarter person than me is perhaps able to hack an eBPF HID program together to temporarily remediate the problem
oakwhiz 3y ago
It kind of sounds like there needs to be a quirks mode for pens.
abdullahkhalids 3y ago
I recently started a remote scientist job, and its kind of insane that I am the only person among dozens who use a tablet+stylus during meetings. People have such poor discussions, because unlike a irl meeting where multiple people will write stuff on a whiteboard, here people are just communicating verbally (read vaguely) or painfully scratching unintelligible diagrams/equations with their mouse. I think every remote job that sends employees a laptop should also be sending them a wacom tablet.
- jon-wood 3y ago
  What are you using for note taking? I while back I bought a small Wacom tablet for this very purpose, particularly being able to quickly sketch out diagrams while in in meetings, but ended up giving up because I couldn't get the hang of using the tablet. I'm pretty sure it was a tooling issue rather than the tablet itself, and wouldn't mind taking another punt at it.
  - orbital-decay 3y ago
    What's your OS? On Windows, Nebo is pretty good. Rnote is the best of FOSS, but it lacks handwriting recognition if you need it, at least for now. For hybrid workflows (tablet + keyboard) Obsidian also has something to offer, Excalidraw plugin in particular. >I'm pretty sure it was a tooling issue Depends on whether you want shape/diagram recognition or not. If the only thing you need is free-form drawing, you can use just about anything.
    - abdullahkhalids 3y ago
      I just tried Rnote. Seems quite decent. But on MacOS, it seems to be bugged. Instead of a tiny pointer/eraser/other tool showing as the pointer, it continues to show the OS mouse pointer. I will try it on linux later. Hopefully it will work there. My initial gripe is that there are three different toolbars, all separated from each other. I want to change tools and their settings quickly, not move my hand across the entire screen. For example, select brush tool on bottom toolbar, select size on the left toolbar, then select color on the top toolbar.
pengaru 3y ago
Maybe this can be worked around via the 'quirks=' usbhid module parameter: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/hid/usbhid/hid-core.c?h=v6.5#n64 https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin... $ modinfo -p usbhid | grep quirks quirks:Add/modify USB HID quirks by specifying quirks=vendorID:productID:quirks where vendorID, productID, and quirks are all in 0x-prefixed hex (array of charp)
lantastic 3y ago
David reached out to the maintainers on the relevant ML [0]. Could also CC the regressions@lists.linux.dev [1]. [0] https://lore.kernel.org/linux-input/nycvar.YFH.7.76.2311012033290.29220@cbobk.fhfr.pm/ https://lore.kernel.org/linux-input/nycvar.YFH.7.76.23110120... [1] https://www.kernel.org/doc/html/v6.6/admin-guide/reporting-regressions.html https://www.kernel.org/doc/html/v6.6/admin-guide/reporting-r...
- molticrystal 3y ago
  It didn't hurt that the comment section[0] for the blog is powered by the fediverse [1], so Greg KH responded to the post [2] pointing him towards writing that email and tips on doing so along with many other tech savvy individuals. This issue is being seen all over the place. [0] https://www.davidrevoy.com/article995/how-a-kernel-update-broke-my-stylus-need-help/show#comments https://www.davidrevoy.com/article995/how-a-kernel-update-br... [1] https://framapiaf.org/@davidrevoy/111336038253784524 https://framapiaf.org/@davidrevoy/111336038253784524 [2] https://social.kernel.org/notice/AbN55QfONFCPweYq7E https://social.kernel.org/notice/AbN55QfONFCPweYq7E
- charcircuit 3y ago
  >No need to file some ticket in some random web form with some odd account, email. It's simple, and easy (note, turn off HTML mode for it please.) Does he not realize how poor of a user experience it is to interact with mailing lists compared to actual bug trackers?
  - db48x 3y ago
    It’s email. How is it a poor experience? You literally just type out a description of your problem into your favorite email program, one you have used thousands of times in your life, put the address of the mailing list in the To field, and send it. Then when someone replies, you get the reply as an email. If that’s a poor user experience for you, then you are using the wrong email program.
    - whensdinner 3y ago
chabad360 3y ago
While very annoying, it is possible to create an evdev based wrapper around the input, that can remap any of the signals as you wish (I know it's possible cause I've done it).
- jeroenhd 3y ago
  According to the evtest screenshots the author added (https://www.davidrevoy.com/data/images/blog/2023/2023-11-01_linux-kernel-broke-my-stylus_evtest-compare.jpg https://www.davidrevoy.com/data/images/blog/2023/2023-11-01_...) that may need a rather annoying and unreliable wrapper. The kernel transmits the buttons as BTN_TOOL_RUBBER, as far as I can see.
gertlex 3y ago
We had a fun one a couple years ago where a kernel bump turned our USB touchscreen monitors into trackpad monitors... (i.e. touch input behaved like your laptop trackpad) Thankfully I was able to figure out a udev rule via trial and error and get that fixed easily enough. ChatGPT probably would have saved me some time there had I had it at the time. Deploying that to hardware in the field was the messy bit (but still automated, so that was a win).
1970-01-01 3y ago
Great example of how Linux evenings randomly come about. Someone, somewhere, did a thing, and now user needs to find out who did what, and why it happend, and how to fix it by themselves.
- ploum 3y ago
  While with proprietary software, someone, somewhere, did a thing and now you need to change your whole workflow because it is the new "only way to do it". But you will like it because marketing said so.
  - ketralnis 3y ago
    Absolutely. I may not be happy about the people that do the free work for me doing work I wish they didn't do. But at least I'm sufficiently in control that I can roll back, or that if literally the whole world hates the change that it can be forked. Whereas Excel had the bug with date format detection that the entire world had to work around with no chance of getting it fixed. The genomics field had to change the name of a gene because even though they all knew it and must have had some population of people that had the skills to fix it, just couldn't until Microsoft got sufficiently embarassed by it after decades (https://gizmodo.com/microsoft-fixes-excel-feature-that-forced-scientists-to-1850949443 https://gizmodo.com/microsoft-fixes-excel-feature-that-force...)
  - numpad0 3y ago
    Everyone frantically rushes to archived versions of Wacom drivers and starts installing random .reg files from the Internet when Microsoft do this(they did couple times).
  - pjmlp 3y ago
    Still less painfull than spending hours researching to find something that works, only to break at any kernel update. Got my first GNU/Linux in 1995 with Slackware 2.0, kernel 1.0.9, got fed up by the time Windows 7 got released, still own a Asus 1215B Netbook with Ubuntu, with its own share of kernel updates breaking graphics and wlan. Anything else, happy with Samsung, LG, Nokia, Apple, Google and Microsoft overlords.
Galanwe 3y ago
If it's just a driver problem, why not checkout the old one, build it as a new module with a different ID, and insmod it on the new kernel?
- dharmab 3y ago
  Because an artist using, say, Krita on Linux to paint shouldn't be expected to rebuild kernel modules.
  - doublerabbit 3y ago
    > Krita on Linux to paint shouldn't be expected to rebuild kernel modules. The other-side of the coin is that if your using Linux, you should expect that such things happen without your desire. And that if you wish for the previous feature you'll have to nose-dive in to if you want to fix or just revert back to the previous version. You shouldn't, but here we are. At the risk of downvotes, I would also say you shouldn't really need to update the kernel either. Unless there is really a feature required, or extreme-security vulnerability of your current version, exclude it. That was the past-beauty of Linux, you didn't need to be on $latest; that was Microsoft's realm, always requiring updates. If it worked you left it be for this exact reason. Every day I look at my iPhone and everyday there's a batch of updates for my apps sometimes twice on the same day for the same app and I don't use many. The current fearing update culture we live in is terrible. /vent
    - pests 3y ago
      > if your using Linux, you should expect that such things happen without your desire Why?
      - doublerabbit 3y ago
        Because Linux is a community ran, free. You can't demand a volenteer maintainer to keep it the way you like it. You either accept it, keep back/change or incentive the maintainer. If using the second button as a eraser is the de-facto feature, then it's likely it's going to be implemented the same everywhere else. Or could simply be the maintainer doesn't like his art work, a sadist and changed for the sake of pure evil. Either way you can't expect anything to stay as you like it if it's free. It's why forks/distributions exist.
jdhendrickson 3y ago
I love his comic, I hope this is addressed soon.
qwery 3y ago
Software all the way up the input stack seems to be a bit confused about a few things, or there is a lack of understanding that: - graphics tablet digitisers generally are not and don't include touch digitisers - tablets are not touch screens - a stylus (tip) isn't a button - a stylus can have buttons By convention, stylus tip interactions are "tools", which seems to be what Linux' BTN_TOOL_* codes are meant to mean. A button on a stylus should never be (directly) translated to a tool -- like how the ~5th button on a mouse is `BTN_EXTRA`, not `BTN_MOUSE`. In this particular case it looks like the thinking is: the device has 2 event codes (which we know as the stylus tip and the barrel button) and it's a stylus^wdigitiser, so map device[0] to `BTN_DIGI` and device[1] to `BTN_DIGI + 1`. editendum: the thing about the BTN_EXTRA doesn't really map... The "digitiser" event codes are being treated in the same way as mouse event codes (device[i] => event[CLASS + i]). The issue is really that the barrel buttons have no (dedicated) event codes.
- excircul 3y ago
  The XP-Pen hardware reports HID "Eraser" usage (some digitizers with accompanying HID "Invert" usage, others without) for the top button. Linux' generic driver does here nothing more than forwarding the corresponding events from hardware to the userspace. The missing Invert usage caused Linux to report Eraser as regular tip touch to the userspace, which was not intended but caused by an oversight in the kernel code. This, together with a bunch of hacks (Digimend and X11 tweaks), allowed OP to map that button to a right click. Now, with the kernel logic fixed, these hacks do not work anymore. This is a case of xkcd://1172. The userspace solution is to remap BTN_TOOL_RUBBER to a right click for the weird stylus instead.
  - qwery 3y ago
    Yes, barrel switches should ideally be reported as "Barrel Switch" and "Secondary Barrel Switch". I have a Huion tablet[-1] that reports the second barrel switch as Eraser. The secondary barrel switch was not in HID originally[0], and I guess some devices were designed to follow an earlier spec -- that, or avoiding Wacom patents. [-1] I don't remember if it's also a uclogic controller? [0] https://www.usb.org/sites/default/files/hutrr46e.txt https://www.usb.org/sites/default/files/hutrr46e.txt
javier_e06 3y ago
Oh don't get me started. I got a Corsair Claw Gaming mouse and Rocky Linux 9 does not have a simple way to map the buttons to events. Corsair does not provide Linux software for it. Yes I tried all kind of mapping programs to no avail. My chromebook old chromebook tablet (32 bit) has a stylus and works okay. I won't upgrade, I know how those upgrades break non-run-of-the-mill external devices APIs.
- jwrallie 3y ago
  OK, I will byte, have you tried Input Remapper[1] ? I think most of your problem is coming from using Rocky Linux, it is much easier to find support on Ubuntu derivatives for tools made by the community in general. [1] https://github.com/sezanzeb/input-remapper https://github.com/sezanzeb/input-remapper
  - javier_e06 3y ago
    Thanks for the lead. No dnf package for Rocky Linux for input-remapper Tried to build it manually and bailed out with error: error: can't find Rust compiler That is usually the beginning of rabbit-hole.
paxys 3y ago
Ah, Linux.
pipeline_peak 3y ago
Definitely wouldn’t recommend using Linux for artwork…
dfc 3y ago
I wish the digimend kernel modules could get mainlined. It kills me that you need to install out of tree modules in order to use a graphics tablet.
raincole 3y ago
Stylus is like printer. It has existed since forever, but somehows it's always broken randomly.
numpad0 3y ago
OT, but PSA: If anyone has issues with tablet PC screens feeling slimy and uncomfortable to draw on, try "paperlike" screen protectors. It has grainy finishing reminiscent of tracing papers and office desk surfaces, and significantly improves drawing feel.
Rapzid 3y ago
Interesting. Should probably be mappable.. But, having to flip a stylus around just to engage with the eraser functionality in the app is perhaps an outmoded concept. The author makes it sound like the fault out-of-touch stylus-lite business users, but I disagree. Some artists erase a lot as part of their workflow, particularly during certain parts, and map this functionality accordingly..
baz00 3y ago
Yeah this is why I don't use Linux on the desktop. 25 years of trying and at least once every 6 months something stops working completely or doesn't work out of the box properly. I gave up.
- prmoustache 3y ago
  Nah, only when wanting to use exotic hardware/software that is not supported by the vendor. The article author wouldn't have had this issue with a wacom.
  - baz00 3y ago
    I have those issues with power management and graphics drivers on stock Intel machines. The real problem is the kernel QA is shite and the vendor packaging QA is shite. Microsoft are actually orders of magnitude better and that's saying something.
    - prmoustache 3y ago
      I don't know. My professional laptop on windows decided one day to only discover (or not) randomly the external monitors connected to my usb-c dock (same manufacturer as laptop). If I am booting on windows, every time I boot it is like rolling a dice to know if my screens will show something. Do not happen on Linux on same laptop. So I would say you have rose tinted glasses when using windows or just got lucky with it and unlucky with linux.
  - d3w4s9 3y ago
    "not supported by the vendor" I'd say you probably should think about it the other way -- you only get decent Linux desktop experience if everything is supported by the vendor. Just look up on HN what kind of trouble people run into when installing a Linux distro on their laptop. People go as far as recommending very specific models from certain brands because everything works as expected on those laptops.
    - vkazanov 3y ago
zhewang87 3y ago
[dead]

213 points by yorwba 3y ago

thanatos519 3y ago

Sounds like someone broke userland!

Aleklart 3y ago
“No, we fixed the glitch.”
- tux3 3y ago
  That doesn't normally fly in kernel land If a 'fix' broke something (and it's not for something like a major security issue), then it is customary to revert the fix and go back to the drawing board I can see that this is a complicated situation and OP was a special case that had not been affected by acccident, but those are excuses If you upgrade your kernel and you don't have any out of tree stuff, by policy nothing should break, orherwise it's the kernel's problem.
  - smelendez 3y ago
    > go back to the drawing board I see what you did there
  - appleskeptic 3y ago
    “Never break userland” is more of a strong guideline than a hard rule, even if Linus has sometimes made it sound otherwise during a heated gamer moment. The fact is, kernel devs break userland code all the time, sometimes because the alternative is worse, or sometimes because a bugfix broke something and no one noticed until much later and it was decided not to revert the fix because it would now break other things to do that. Try to run an entire large userland from 2001 on a modern kernel and tell me how that goes.
    - tux3 3y ago
      Mistakes are human. Certainly a lot of bugs slip through, and that's understandable, software is hard. That being said, I find that people tend to take the no regression rule pretty seriously, if you bring a valid complaint. It makes good sense. We don't want to give people a reason to stay on old kernels even more than they do today (this post written on an ancient Android kernel, chock full of out of tree drivers). We can at least save people from pain when it comes to in-tree drivers. I think we really all benefit from that policy. Even if it sometimes keeps some less than perfect workarounds in the code
    - dfc 3y ago
      I always assumed it was a rule. I don't know of any kernel regressions that broke something I used daily. Are there any examples you can think of?
  - jeroenhd 3y ago
    I mean, for whatever stupid and obscure reason, the manufacturer decided to hard-code the eraser button like that. The device was designed to have two erasers. If you used that tablet on Windows and switched to Linux, you'd notice the device wasn't working like it was supposed to. I can see why one would consider that a bug. One user asks "why is my right click an eraser now", another asks "why is my eraser a right click on Linux". Which one is right? To resolve this, the devs could've added an easy setting somewhere to control this behaviour, but I still think this can be considered a reasonable bugfix as the manufacturer decided on this stupid design. Linux drivers used to hack around that, but that's a classic XKCD 1172 problem.
tedunangst 3y ago
Every userland program is equal, but some are more equal than others.

kaetemi 3y ago

Now Windows isn't alone in changing or resetting drawing tablet settings on random updates.

DarkmSparks 3y ago
difference with Linux is this will be fixed most likely before new year. windows will always stay broken.
- m463 3y ago
  windows may be fixed (but the update will also fix your default browser to be edge)
- pjmlp 3y ago
  Depends, my wlan driver on Ubuntu still has connection issues with my router, it is the only device in the house that requires being occasionally plugged into LAN cable. It is also the only one using GNU/Linux, the TV/WebOS and Android/Linux devices, alongside Windows, just work.

pizzapim 3y ago

A smarter person than me is perhaps able to hack an eBPF HID program together to temporarily remediate the problem

oakwhiz 3y ago

It kind of sounds like there needs to be a quirks mode for pens.

abdullahkhalids 3y ago

I recently started a remote scientist job, and its kind of insane that I am the only person among dozens who use a tablet+stylus during meetings. People have such poor discussions, because unlike a irl meeting where multiple people will write stuff on a whiteboard, here people are just communicating verbally (read vaguely) or painfully scratching unintelligible diagrams/equations with their mouse. I think every remote job that sends employees a laptop should also be sending them a wacom tablet.

jon-wood 3y ago
What are you using for note taking? I while back I bought a small Wacom tablet for this very purpose, particularly being able to quickly sketch out diagrams while in in meetings, but ended up giving up because I couldn't get the hang of using the tablet. I'm pretty sure it was a tooling issue rather than the tablet itself, and wouldn't mind taking another punt at it.
- orbital-decay 3y ago
  What's your OS? On Windows, Nebo is pretty good. Rnote is the best of FOSS, but it lacks handwriting recognition if you need it, at least for now. For hybrid workflows (tablet + keyboard) Obsidian also has something to offer, Excalidraw plugin in particular. >I'm pretty sure it was a tooling issue Depends on whether you want shape/diagram recognition or not. If the only thing you need is free-form drawing, you can use just about anything.
  - abdullahkhalids 3y ago
    I just tried Rnote. Seems quite decent. But on MacOS, it seems to be bugged. Instead of a tiny pointer/eraser/other tool showing as the pointer, it continues to show the OS mouse pointer. I will try it on linux later. Hopefully it will work there. My initial gripe is that there are three different toolbars, all separated from each other. I want to change tools and their settings quickly, not move my hand across the entire screen. For example, select brush tool on bottom toolbar, select size on the left toolbar, then select color on the top toolbar.

pengaru 3y ago

Maybe this can be worked around via the 'quirks=' usbhid module parameter: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/hid/usbhid/hid-core.c?h=v6.5#n64 https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin... $ modinfo -p usbhid | grep quirks quirks:Add/modify USB HID quirks by specifying quirks=vendorID:productID:quirks where vendorID, productID, and quirks are all in 0x-prefixed hex (array of charp)

lantastic 3y ago

David reached out to the maintainers on the relevant ML [0]. Could also CC the regressions@lists.linux.dev [1]. [0] https://lore.kernel.org/linux-input/nycvar.YFH.7.76.2311012033290.29220@cbobk.fhfr.pm/ https://lore.kernel.org/linux-input/nycvar.YFH.7.76.23110120... [1] https://www.kernel.org/doc/html/v6.6/admin-guide/reporting-regressions.html https://www.kernel.org/doc/html/v6.6/admin-guide/reporting-r...

molticrystal 3y ago
It didn't hurt that the comment section[0] for the blog is powered by the fediverse [1], so Greg KH responded to the post [2] pointing him towards writing that email and tips on doing so along with many other tech savvy individuals. This issue is being seen all over the place. [0] https://www.davidrevoy.com/article995/how-a-kernel-update-broke-my-stylus-need-help/show#comments https://www.davidrevoy.com/article995/how-a-kernel-update-br... [1] https://framapiaf.org/@davidrevoy/111336038253784524 https://framapiaf.org/@davidrevoy/111336038253784524 [2] https://social.kernel.org/notice/AbN55QfONFCPweYq7E https://social.kernel.org/notice/AbN55QfONFCPweYq7E
charcircuit 3y ago
>No need to file some ticket in some random web form with some odd account, email. It's simple, and easy (note, turn off HTML mode for it please.) Does he not realize how poor of a user experience it is to interact with mailing lists compared to actual bug trackers?
- db48x 3y ago
  It’s email. How is it a poor experience? You literally just type out a description of your problem into your favorite email program, one you have used thousands of times in your life, put the address of the mailing list in the To field, and send it. Then when someone replies, you get the reply as an email. If that’s a poor user experience for you, then you are using the wrong email program.
  - whensdinner 3y ago

chabad360 3y ago

While very annoying, it is possible to create an evdev based wrapper around the input, that can remap any of the signals as you wish (I know it's possible cause I've done it).

jeroenhd 3y ago
According to the evtest screenshots the author added (https://www.davidrevoy.com/data/images/blog/2023/2023-11-01_linux-kernel-broke-my-stylus_evtest-compare.jpg https://www.davidrevoy.com/data/images/blog/2023/2023-11-01_...) that may need a rather annoying and unreliable wrapper. The kernel transmits the buttons as BTN_TOOL_RUBBER, as far as I can see.

gertlex 3y ago

We had a fun one a couple years ago where a kernel bump turned our USB touchscreen monitors into trackpad monitors... (i.e. touch input behaved like your laptop trackpad) Thankfully I was able to figure out a udev rule via trial and error and get that fixed easily enough. ChatGPT probably would have saved me some time there had I had it at the time. Deploying that to hardware in the field was the messy bit (but still automated, so that was a win).

1970-01-01 3y ago

Great example of how Linux evenings randomly come about. Someone, somewhere, did a thing, and now user needs to find out who did what, and why it happend, and how to fix it by themselves.

ploum 3y ago
While with proprietary software, someone, somewhere, did a thing and now you need to change your whole workflow because it is the new "only way to do it". But you will like it because marketing said so.
- ketralnis 3y ago
  Absolutely. I may not be happy about the people that do the free work for me doing work I wish they didn't do. But at least I'm sufficiently in control that I can roll back, or that if literally the whole world hates the change that it can be forked. Whereas Excel had the bug with date format detection that the entire world had to work around with no chance of getting it fixed. The genomics field had to change the name of a gene because even though they all knew it and must have had some population of people that had the skills to fix it, just couldn't until Microsoft got sufficiently embarassed by it after decades (https://gizmodo.com/microsoft-fixes-excel-feature-that-forced-scientists-to-1850949443 https://gizmodo.com/microsoft-fixes-excel-feature-that-force...)
- numpad0 3y ago
  Everyone frantically rushes to archived versions of Wacom drivers and starts installing random .reg files from the Internet when Microsoft do this(they did couple times).
- pjmlp 3y ago
  Still less painfull than spending hours researching to find something that works, only to break at any kernel update. Got my first GNU/Linux in 1995 with Slackware 2.0, kernel 1.0.9, got fed up by the time Windows 7 got released, still own a Asus 1215B Netbook with Ubuntu, with its own share of kernel updates breaking graphics and wlan. Anything else, happy with Samsung, LG, Nokia, Apple, Google and Microsoft overlords.

Galanwe 3y ago

If it's just a driver problem, why not checkout the old one, build it as a new module with a different ID, and insmod it on the new kernel?

dharmab 3y ago
Because an artist using, say, Krita on Linux to paint shouldn't be expected to rebuild kernel modules.
- doublerabbit 3y ago
  > Krita on Linux to paint shouldn't be expected to rebuild kernel modules. The other-side of the coin is that if your using Linux, you should expect that such things happen without your desire. And that if you wish for the previous feature you'll have to nose-dive in to if you want to fix or just revert back to the previous version. You shouldn't, but here we are. At the risk of downvotes, I would also say you shouldn't really need to update the kernel either. Unless there is really a feature required, or extreme-security vulnerability of your current version, exclude it. That was the past-beauty of Linux, you didn't need to be on $latest; that was Microsoft's realm, always requiring updates. If it worked you left it be for this exact reason. Every day I look at my iPhone and everyday there's a batch of updates for my apps sometimes twice on the same day for the same app and I don't use many. The current fearing update culture we live in is terrible. /vent
  - pests 3y ago
    > if your using Linux, you should expect that such things happen without your desire Why?
    - doublerabbit 3y ago
      Because Linux is a community ran, free. You can't demand a volenteer maintainer to keep it the way you like it. You either accept it, keep back/change or incentive the maintainer. If using the second button as a eraser is the de-facto feature, then it's likely it's going to be implemented the same everywhere else. Or could simply be the maintainer doesn't like his art work, a sadist and changed for the sake of pure evil. Either way you can't expect anything to stay as you like it if it's free. It's why forks/distributions exist.

jdhendrickson 3y ago

I love his comic, I hope this is addressed soon.

qwery 3y ago

Software all the way up the input stack seems to be a bit confused about a few things, or there is a lack of understanding that: - graphics tablet digitisers generally are not and don't include touch digitisers - tablets are not touch screens - a stylus (tip) isn't a button - a stylus can have buttons By convention, stylus tip interactions are "tools", which seems to be what Linux' BTN_TOOL_* codes are meant to mean. A button on a stylus should never be (directly) translated to a tool -- like how the ~5th button on a mouse is `BTN_EXTRA`, not `BTN_MOUSE`. In this particular case it looks like the thinking is: the device has 2 event codes (which we know as the stylus tip and the barrel button) and it's a stylus^wdigitiser, so map device[0] to `BTN_DIGI` and device[1] to `BTN_DIGI + 1`. editendum: the thing about the BTN_EXTRA doesn't really map... The "digitiser" event codes are being treated in the same way as mouse event codes (device[i] => event[CLASS + i]). The issue is really that the barrel buttons have no (dedicated) event codes.

excircul 3y ago
The XP-Pen hardware reports HID "Eraser" usage (some digitizers with accompanying HID "Invert" usage, others without) for the top button. Linux' generic driver does here nothing more than forwarding the corresponding events from hardware to the userspace. The missing Invert usage caused Linux to report Eraser as regular tip touch to the userspace, which was not intended but caused by an oversight in the kernel code. This, together with a bunch of hacks (Digimend and X11 tweaks), allowed OP to map that button to a right click. Now, with the kernel logic fixed, these hacks do not work anymore. This is a case of xkcd://1172. The userspace solution is to remap BTN_TOOL_RUBBER to a right click for the weird stylus instead.
- qwery 3y ago
  Yes, barrel switches should ideally be reported as "Barrel Switch" and "Secondary Barrel Switch". I have a Huion tablet[-1] that reports the second barrel switch as Eraser. The secondary barrel switch was not in HID originally[0], and I guess some devices were designed to follow an earlier spec -- that, or avoiding Wacom patents. [-1] I don't remember if it's also a uclogic controller? [0] https://www.usb.org/sites/default/files/hutrr46e.txt https://www.usb.org/sites/default/files/hutrr46e.txt

javier_e06 3y ago

Oh don't get me started. I got a Corsair Claw Gaming mouse and Rocky Linux 9 does not have a simple way to map the buttons to events. Corsair does not provide Linux software for it. Yes I tried all kind of mapping programs to no avail. My chromebook old chromebook tablet (32 bit) has a stylus and works okay. I won't upgrade, I know how those upgrades break non-run-of-the-mill external devices APIs.

jwrallie 3y ago
OK, I will byte, have you tried Input Remapper[1] ? I think most of your problem is coming from using Rocky Linux, it is much easier to find support on Ubuntu derivatives for tools made by the community in general. [1] https://github.com/sezanzeb/input-remapper https://github.com/sezanzeb/input-remapper
- javier_e06 3y ago
  Thanks for the lead. No dnf package for Rocky Linux for input-remapper Tried to build it manually and bailed out with error: error: can't find Rust compiler That is usually the beginning of rabbit-hole.

paxys 3y ago

Ah, Linux.

pipeline_peak 3y ago

Definitely wouldn’t recommend using Linux for artwork…

dfc 3y ago

I wish the digimend kernel modules could get mainlined. It kills me that you need to install out of tree modules in order to use a graphics tablet.

raincole 3y ago

Stylus is like printer. It has existed since forever, but somehows it's always broken randomly.

numpad0 3y ago

OT, but PSA: If anyone has issues with tablet PC screens feeling slimy and uncomfortable to draw on, try "paperlike" screen protectors. It has grainy finishing reminiscent of tracing papers and office desk surfaces, and significantly improves drawing feel.

Rapzid 3y ago

Interesting. Should probably be mappable.. But, having to flip a stylus around just to engage with the eraser functionality in the app is perhaps an outmoded concept. The author makes it sound like the fault out-of-touch stylus-lite business users, but I disagree. Some artists erase a lot as part of their workflow, particularly during certain parts, and map this functionality accordingly..

baz00 3y ago

Yeah this is why I don't use Linux on the desktop. 25 years of trying and at least once every 6 months something stops working completely or doesn't work out of the box properly. I gave up.

prmoustache 3y ago
Nah, only when wanting to use exotic hardware/software that is not supported by the vendor. The article author wouldn't have had this issue with a wacom.
- baz00 3y ago
  I have those issues with power management and graphics drivers on stock Intel machines. The real problem is the kernel QA is shite and the vendor packaging QA is shite. Microsoft are actually orders of magnitude better and that's saying something.
  - prmoustache 3y ago
    I don't know. My professional laptop on windows decided one day to only discover (or not) randomly the external monitors connected to my usb-c dock (same manufacturer as laptop). If I am booting on windows, every time I boot it is like rolling a dice to know if my screens will show something. Do not happen on Linux on same laptop. So I would say you have rose tinted glasses when using windows or just got lucky with it and unlucky with linux.
- d3w4s9 3y ago
  "not supported by the vendor" I'd say you probably should think about it the other way -- you only get decent Linux desktop experience if everything is supported by the vendor. Just look up on HN what kind of trouble people run into when installing a Linux distro on their laptop. People go as far as recommending very specific models from certain brands because everything works as expected on those laptops.
  - vkazanov 3y ago

zhewang87 3y ago

[dead]