Re: [Off Topic Conspiracy Theories] RE: UDI and Free(tm) Software

david parsons (o.r.c@p.e.l.l.p.o.r.t.l.a.n.d.o.r.u.s)
8 Oct 1998 00:20:48 -0700


In article <linux.kernel.3632415c.14311129@smtp.tm.net.my>,
<teamwork@freemail.c3.hu> wrote:
>David said:
>
> "This brings up another (also off-topic, sorry)
> advantage of coding to a UDI interface. If you
> can't get the vendors driver to work on your
> system and they've coded to a UDI interface,
> it's a whole bunch easier to take a
> source-available driver from another system and
> put it into place to deal with the bugs."
>
>
>I am too dense, so please educate me on this, and see if my feeble attempt on
>logic makes any sense:
>
>
>Let A = Vendor's driver working on your system.
>
>Let B = Source-available driver from another system.
>
>Let C = Working driver (source) on your system.
>
>
>What you said essentially is [*]
>
> I. [not A] = [B],
>
> and II. [B] = [C],
>
>therefore [not A] somehow leads to [C], which to you, is satisfactory.

I can't follow your, umm, logic.

If I've got a non-Linux machine with a binary-only UDI driver that
doesn't work, and I've got a Linux UDI driver, it's a fairly simple
matter to recompile the driver for that other system and get that
system to its feet.

>Sorry, David, my simple brain just doesn't work this way.
>
> 1. If vendor's binary UDI drivers don't work on your system,
> what make you think the vendor will release source driver?

No, that's not what I said. You're assuming that if Linux has the
UDI interface (which it may, because I've got a whole wad of free
time coming up and need a break from Mastodon; I've got enough
machines so I won't mind much if I smoke one) that every Linux
driver writer will just give up and let people use the drivers from
vendors, whether or not source is available. This is bogus, because
writing device drivers is fun, and writing to a fixed published
interface doesn't make in any less fun.

If my SCO or NT box has a broken driver and that driver is written
to the UDI interface, I can take the freely distributable driver
from Linux[1], recompile it on that system, and save my little tushie.

>A side effect of UDI is that encourages the vendors to distribute only binary
>drivers,

No it doesn't. On the contrary, it encourages the vendors to
release the sources because there are a lot of Unices out there, and
no vendor can support a test network for all of them, but if the
Unices adhere to the UDI spec, you can build and test your device
driver on ONE system, put the sources up on your website, and tell
customer for obscure Unix xyzzy that all they need to do is download
zipchip(tm).tar.gz and follow the install instructions.

That's the nice thing about standards; standards encourage publishing
sources, because you don't have to worry about getting bitten badly
by some bozo improperly hacking the driver and eating 30 billion
transaction records two hours before the nightly backup.

>BTW, David, I have read many of your message regarding UDI and I find that your
>support for UDI have been way too unreasonably giddy.

That's okay; I've found, with the rare exception of the technical
complaints, that the opposition for UDI could have been taken
directly from an episode of the X-Files.

____
david parsons \bi/ "What, you hacked the UDI driver and lost all those
\/ records? Bummer for you, because the license says
you patch, you lose."

[1: I'm assuming, reasonably, that there WILL be a linux driver,
because I don't buy hardware that doesn't come with specs.
People who buy the systems I build may not adhere to that logic
and UDI wins there too, but that's a different story.]

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/