Fedora

Bug #1 (liberation)
Comment #1248

Comment 1248 for bug 1

Revision history for this message

Faldegast (faldegast) wrote on 2010-07-06: RE: [Bug 1] Microsoft has a majority market share

#1248

> Date: Mon, 5 Jul 2010 22:08:17 +0000
> From: <email address hidden>
> To: <email address hidden>
> Subject: [Bug 1] Microsoft has a majority market share
>
> Hi there.
>
>
> > I dont see how you get from portable binary drivers to binary blobs.
>
> If you factor in my understanding of the word "blob" (which for me
> equals "binary (only) drivers", not "free drivers in their pre-compiled
> form"), then you just answered yourself.

I still dont see how that becomes an issue in this case. I am not talking about binary only drivers. However even with binary only drivers it would be far better if they worked with more then one operating system. Personally i never use binary only drivers. I maintain many machines using repositories for upgrades, so using something that is not shipped by the distribution is not an option. Stuff like Adobe Flash is ok, if it fails it breaks nothing else. If the ATI drivers fail the screen goes black...

> > With my suggestion you would be able to for example use the GPL
> drivers from Linux on FreeBSD or Solaris without recompiling them. If we
> also would be able to agree on standard for packages and repositories
> you would also be able to just install the drivers from a common
> repository.
>
> And that's what I agreed would be really cool about this universal
> driver interface (among other things).

It is the main thing that i think will make the selection of OS:es about something else then hardware compatibility.

> Anyway, I've been thinking about this a little more and here's the
> result:
> Got a flash drive, external hard drive, memory card reader, PostScript
> printer or a decent webcam embedded to your laptop (just a few examples
> from the top of my head)? They all work like a charm right out of the
> box (and they probably did long before you first decided to try some
> random free OS), right? That's because all these devices adhere to open
> standards defining how particular sort of hardware has to behave from
> the software perspective - both input and output is well defined. They
> must understand certain commands and be able to at least explain their
> output to the driver if it's mandatory format isn't already defined
> within the standard itself. That way every device in the same hardware
> class becomes just a different hardware implementation of the same
> concept with SIMPLE GENERIC DRIVERS (one driver to rule them all) and
> YOU GET A FLAWLESS OUT OF THE BOX SUPPORT FOR YOUR DEVICE. I think that
> what we really need is to put more pressure on hardware manufacturers to
> make them move in that direction which combined with your idea would
> blast all the hardware support M$ likes to brag about so much into
> oblivion.
For printers then we have the standard CUPS, for which drivers work in a platform-independent way. This is userspace drivers which printer drivers has traditionally been in *nix. When the printers use PS and can stick with the generic CUPS PS driver it is extra beautiful.
Flash drives and external hard drives are specified in the USB specification. This kind of generic interfaces is how everything should work. However even if the USB drive can be used with a generic driver, the USB port itself is not generic. There is only a few vendors of USB IP designs but still we need more then one USB driver.
I do not really agree that this is flawless. In many cases the binary blobs have moved to the hardware device, where we do not have any ability to patch it should it contain bugs. PostScript printers is a good example. Some printers have firmware that are quite buggy making them really hard to use. Had their software been thinner and the main work done in the main system we would have an easier job patching it. Sometimes these problems are adressed with firmware updates, that potentially can render the device useless.

But there are still ways to fix this with generic drivers, and that is to create standards for how simple devices can be made. Here the ATA and SATA standard are a good example. It is based on many subset of commands. For each subsets all commands are mandatory, but suppliers can choose what subsets to implement. Apply this for printers and we cover even the cheap printers that does not have the processing power to run postscript.

Putting pressure on hardware manufacturers is a good idea. There should be an OSS compliance list where we could find companies that support open drivers. And that is compliance, not compatibility. I dont care much for hardware suppliers that does not contribute, even if there often are good OSS drivers i feel more comfortable buying something with "Linux" printed on the box...

As nerds we have a responsibility here. Its to us people turn when they have computer problems. For the last few years my policy have been that i only support Linux, i dont answer any Windows questions even if i know them. I simply reply "I dont use Windows". As a result the amount of time i have to spend supporting friends and family have decreased dramatically. Most of them switched to Fedora or Ubuntu. I probably dont have to tell you how much easier it is support these operating systems.

_________________________________________________________________
Är din dator en skvallerbytta? Testa den här!
http://channels.se.msn.com/channelizers/IE_skvallerbyttan.aspx

> Date: Mon, 5 Jul 2010 22:08:17 +0000
> From: JnS.002@gmail.com
> To: faldegast@hotmail.com
> Subject: [Bug 1] Microsoft has a majority market share
> 
> Hi there.
> 
> 
> > I dont see how you get from portable binary drivers to binary blobs.
> 
> If you factor in my understanding of the word "blob" (which for me
> equals "binary (only) drivers", not "free drivers in their pre-compiled
> form"), then you just answered yourself.

> > With my suggestion you would be able to for example use the GPL
> drivers from Linux on FreeBSD or Solaris without recompiling them. If we
> also would be able to agree on standard for packages and repositories
> you would also be able to just install the drivers from a common
> repository.
> 
> And that's what I agreed would be really cool about this universal
> driver interface (among other things).

It is the main thing that i think will make the selection of OS:es about something else then hardware compatibility.
 
> Anyway, I've been thinking about this a little more and here's the
> result:
> Got a flash drive, external hard drive, memory card reader, PostScript
> printer or a decent webcam embedded to your laptop (just a few examples
> from the top of my head)? They all work like a charm right out of the
> box (and they probably did long before you first decided to try some
> random free OS), right? That's because all these devices adhere to open
> standards defining how particular sort of hardware has to behave from
> the software perspective - both input and output is well defined. They
> must understand certain commands and be able to at least explain their
> output to the driver if it's mandatory format isn't already defined
> within the standard itself. That way every device in the same hardware
> class becomes just a different hardware implementation of the same
> concept with SIMPLE GENERIC DRIVERS (one driver to rule them all) and
> YOU GET A FLAWLESS OUT OF THE BOX SUPPORT FOR YOUR DEVICE. I think that
> what we really need is to put more pressure on hardware manufacturers to
> make them move in that direction which combined with your idea would
> blast all the hardware support M$ likes to brag about so much into
> oblivion.
For printers then we have the standard CUPS, for which drivers work in a platform-independent way. This is userspace drivers which printer drivers has traditionally been in *nix. When the printers use PS and can stick with the generic CUPS PS driver it is extra beautiful.
Flash drives and external hard drives are specified in the USB specification. This kind of generic interfaces is how everything should work. However even if the USB drive can be used with a generic driver, the USB port itself is not generic. There is only a few vendors of USB IP designs but still we need more then one USB driver.
I do not really agree that this is flawless. In many cases the binary blobs have moved to the hardware device, where we do not have any ability to patch it should it contain bugs. PostScript printers is a good example. Some printers have firmware that are quite buggy making them really hard to use. Had their software been thinner and the main work done in the main system we would have an easier job patching it. Sometimes these problems are adressed with firmware updates, that potentially can render the device useless.