Hi Miles,
> -----Original Message-----
> From: Miles Lane [mailto:[EMAIL PROTECTED]]
> Sent: Thursday, May 04, 2000 11:51 PM
> To: Alan Cox
> Cc: Tom Rini; Johannes Erdfelt; '[EMAIL PROTECTED]'
> Subject: Re: [linux-usb] USB 2.3.99 jobs (updated)t
>
> Alan Cox wrote:
> >
> > > Ok. So, for the sake of simplicity (same interface and
> whatnot in 2.2 and 2.4)
> > > what's wrong with having to use DevFS in 2.2 as well? It
> does exist, it's still
> >
> > Stability, minimum change, maximum back compatibility. As
> has been pointed
> > out - keeping usbdevfs for 2.2 solves that
>
> I am *just* a tester. But I'd like to offer my two bits.
>
> I understand that the distribution folks are not going to
> overhaul their system installation processes and initialization
> scripts as soon as 2.4 becomes available. Making these changes
> will likely be the biggest chunk of work they'll need to do
> in the process of fully enabling and supporting all the new 2.4
> functionality. I imagine the distributors will want to see long
> beta cycles when they overhaul their setup and initialization scripts.
>
> I wonder, since you (Alan) say that the 2.2 backport will likely
> not make it into the 2.2 tree and since it might make sense for
> the backport to use the usbdevfs approach while the 2.4 tree uses
> another (devfs or some ioctl approach), whether continuing to
> develop the backport makes sense.
I'll give a response without looking ahead at the next 12 emails:
I don't think that we have put any significant time (hours) into
developing the backport patch as a group. Vojtech has done this
almost alone.
Personally I've always been focused on USB delivery for 2.4.
Having the backport patch:
(a) is nice for current 2.2 users who want/need USB or who
can help us test only on 2.2,
(b) is easier on some USB developers,
(c) may be helpful to Linux distributors, but that's not a
traditional Linux kernel goal.
> Personally, I think the most important thing is nailing down
> the implementation for 2.4 and getting a solid 2.4 finished.
> It's pretty disturbing to see how many broad implementation
> features are still being sorted out.
I agree on the 2.4 target. I guess I have to agree on the
broad features issue also since I tried to gather them together
for this thread.
> Am I right in thinking that getting the USB support finished is
> the major hurdle in releasing 2.4?
I'm itching to see Alan's response to this, but I'll wait a few
minutes. From reading (too much of) the lkml, linux-fsdevel,
linux-mm, etc., I know that there are also serious fs/VFS,
memory management, and swap problems in 2.3.99-preN.
If you look at the first paragraph in my message that started
this thread, I hope you can see that I am trying to (a) have
healthy USB support in 2.4 and (b) have it fairly soon [so
that USB is not the gating factor for 2.4].
<quote>
Let's assume (yes, I know what that means) that 2.4.0
will be released soon... what do we need to focus on
in the next few (or many) weeks to assure:
- decent behavior of USB (minimal oops/bug reports)
- satisfied users (at least of devices that we have support for)
</quote>
> Perhaps it would be the most efficient and best use of our
> precious development resources to just make USB in 2.4 solid.
> Folks who have to have USB support would then have a lot of
> incentive to test the USB support and help us make it rock
> solid fast.
>
> Is there some reason I don't know about that will prevent the
> rapid adoption of 2.4 when it is ready (other than the time
> lag associated with the distributions getting new 2.4 releases
> on the market)?
>
> I hope this helps,
> Miles
~Randy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
