Introducing Driver Module Framework | Hacker News Reader

Introducing Driver Module Framework

blogs.windows.com

88 pointsvezycash20 comments

I know it's not really the subject here but I've faced a weird case of "what why ?!?" about a windows driver thingy and it's been on my mind to share it for a couple weeks now

For some weird reason, the package to handle the very generic Synaptics touchpad on both my Asus and MSI laptop was moved out of the driver (+ included config panel to configure it), to a combination of a driver that does stuff but you can't configure, and an obscure windows store app that isn't installed with the driver, and is now the only way to configure it.

Want to disable your touchpad when an usb mouse is connected, remove tapping (quick tap on touch pad = left clic), set up multi touch gesture and whatnot ? Well then you better know that you need that, because there was no indication whatsoever that they removed it from the driver, the panel just disappeared over night (oh and it doesn't keep your old settings either, although in my case it was a plus since that's how I knew something changed).

And windows store search is so miserable that searching for "synaptics" does not bring up the app, and it doesn't show in the windows store "my library" listing.

If anyone is involved in that, I would buy you a beer or three just for the chance to sit in front of you and listen to how that happened (this offer of course extends to anyone on the skype team who can tell me the story of how skype continually gets worse at being a chat app)

This is almost certainly motivated by the push to start selling devices in S-mode (aka, Windows Store-only mode). Though generally-speaking, when you have hardware with Store apps that configure them, Windows is supposed to (and generally does from my experience) automatically install the app, rather than have you look for it.

Having apps running in sandboxes where they only have permission to talk to the device they're built to interface with is more secure, and arguably, the best choice for most non-technical users. I'm glad this is the direction they're moving, as long as the ability for those of us who need it to turn such protections off remains.

Makes sense to me; and I genuinely approve of the intent, but why

- is it not installed when the driver installs (to be clear: if you have the app installed, then in the "mouse properties" control panel you have a new tab that appears called "device settings", it's not an app that you launch on its own, it's not a new control panel, it's at the same spot as before as a tab in a generic windows control panel, there is no placeholder to warn you ... it's a really bad experience)

- was I not warned about it when it moved me from the older/legacy one to that (by windows update)

- can't I find it in windows store search or in my library of installed app (only way to find it to install it that I know of, is to open the direct link in edge and let edge pass it to the windows store app) [if that one is me being an idiot, I will gladly take it; because it's the weirdest one for me]

- did it not carry over my settings (I noticed it because both tapping and disable-if-usb reverted to default settings)

It felt like one of those case where they all played with the "new install" experience and no one took looked at what users doing an update would see.

It's a great move forward, and a great idea, but a terrible user experience. 10 minutes in a beer testing environnement would have told them that.

Microsoft probably had little, if anything to do with the decision. Synaptic put a new driver through the pipeline that ends with it being pushed through Windows Update as part of the normal keeping your drivers updated goal (from Microsoft's perspective). There's a lot of QA that goes into that process , but it's all on the "can this driver take down the system, or only the device it's written for?" It's on Synaptic to make good UX choices for the software they distribute, driver config tools included. And it's on ASUS and Msi to select hardware with stable, quality drivers and config tools.

Getting to sit down and talk with the guys that made those decisions would be pretty great, but it's not Microsoft you're really interested in talking to here.

See this for an overview of the design guidelines--part of which is to break out app functionality: https://docs.microsoft.com/en-us/windows-hardware/drivers/de...

As someone who has suffered on an almost daily basis from the poor quality hardware drivers on the Surface platform anything Microsoft can do to improve driver quality is a win in my book (and would have saved MSFT thousands of dollars as the MSFT store replaced my surface and keyboard multiple times at no charge before I finally tracked the issue down to the Bluetooth stack crashing and taking down the physically connected keyboard, which was happenning to me almost daily)

Reminds me that i have a "fun" one here.

Two tablets, both using the same Intel internals (as best i can tell), but one has a non-functional SD reader while the other works just fine. The difference between them is that one uses 32-bit Windows (no clue why) while the other uses 64-bit windows.

Thing is that the SD slot worked fine when i initially got the tablet, but broke at some point after a major Windows update. And thee is no way i can find to roll back drivers.

I have an older but perfectly functional scanner that is a PITA to get working on Windows because the company never shipped an official 64 bit driver for it. Instead I have to download the driver for a slightly newer model and hack it slightly (add my scanner to the list of devices it supports) to get my scanner to work on Windows. It's so dumb.

As far as I know there isn't a "Postscript over IP" solution for this like there is for printers. There is no generic universally supported protocol that scanners could support that would future proof them.

SANE is a less than completely user friendly experience to set up too. I've found it to be pretty fiddly, although once you get it working it seems to do ok.

Reminds me of some years back when a single driver was added to Linux that allowed it to support 100+ models of webcams. This because every damn brand and model used the same internals, but had different USB IDs.

And we should not forget that RMS' initially rallying call was about printer drivers. where as with the old printer he had the source, so he could add a bit of code to notify the people with jobs in the queue that the printer was out of paper, the new printer only came with a binary blob driver.

Methinks the hardware world is yet another incarnation of what happened in the gpu world, layers of bs folded onto themselves.

A good deal of open hardware, even without sophisticated capabilities would free a lot of time and energy for open source OSes

Same scanner issue here (will try your hack some time)

However >There is no generic universally supported protocol that scanners could support that would future proof them.

There must be... because on Kubuntu it works Plug n Play with the bundled software...

That's SANE, a huge project to re-implement all scanner drivers in one megaproject. It can be a bit of a bear if the plug and play stuff doesn't work. On my scanner it defaults to a driver that only generates a fuzzy black and white image so I have to force it to use the right driver.

Speaking of Windows Update breaking things, my Windows 10 tablet (a cheap Kiano) tends to gain or lose the ability to detect screen orientation every other update. At this point I'm (again) stuck having to manually change the orientation whenever I want to work in portrait tablet mode.

I do not understand how this happens.

This is the framework used to make those crashy surface drivers.

Every time I think about switching from MacOS to Windows I'm glad that I find a post like this that reminds me of how bad Windows is in comparison. I'll happily deal with a slow hardware refresh cycle from Apple if it means things actually work.

Do keep in mind that's a pretty serious selection bias filter you're applying to the tiny fraction of driver complaints people have (and I say that as the guy who kicked off the complaint about drivers)

"Open source" drivers are considerably less useful when they have to be WHQL-signed in order for people to use them.

As someone who has had to do tech support for family members, I can tell you right now that anyone who cannot Google "How to Install Unsigned Drivers in Windows 10" and then follow the directions _does not need to be using an unsigned driver._

Nobody likes the decals about using unsigned drivers on their desktop. You should be able to self-sign, which might also be a bit more secure. Unfortunately, that is not allowed in Windows afaik.

A lot of software does not run with self-signed drivers in use.

Reply on news.ycombinator.com