Hi,
Doug Collinge:
> Is it possible to have two separate chains on one bus? Not that I think
> that's a good idea but whatever interface is chosen should be able to
> represent anything that is possible.
>
That doesn't make sense. You'd have a collision on the conditional read
command that's use
Hello!
Okay, the CVS for OWFS from 7/20/2007 works well enough on a 2.6.18
kernel, as installed, not custom. Also the FUSE version 2.6.1 does
work here. The release tells me that it needs a 2.6 type kernel so its
module is disabled. (Query, why?)
What I've done is select a pair of DS2406 devices,
On 8/2/07, Doug Collinge <[EMAIL PROTECTED]> wrote:
>
> Is it possible to have two separate chains on one bus? Not that I think
> that's a good idea but whatever interface is chosen should be able to
> represent anything that is possible.
>
It appears that normally only 1 chain is possible on a b
Is it possible to have two separate chains on one bus? Not that I think
that's a good idea but whatever interface is chosen should be able to
represent anything that is possible.
On 8/2/07, Paul Alfille <[EMAIL PROTECTED]> wrote:
>
> Regarding symlinks:
>
> owfs can probably do symlinks, but that
Regarding symlinks:
owfs can probably do symlinks, but that might miss the point.
All the entries in OWFS are "virtual" -- generated in ram on the fly. Unless
you specifically want to query the link, it's just as easy to duplicate the
entry.
A bigger problem is that OWFS isn't just a file system
On 8/2/07, Matthias Urlichs <[EMAIL PROTECTED]> wrote:
>
> That would allow people to address the devices by position, which is not
> stable -- you insert another device near the beginning and everything
> still works -- but some time later, when the cache for the ordering
> times out, everything i
Hi,
Jan Kandziora:
> A related question: What about symlinks, so one can see the target node when
> listing "chain" with ls?
That makes a lot of sense.
AFAIK, currently the owfs protocol does not support the concept of a
"symlink". I have no idea how difficult fixing that would be.
--
Matthia
Am Donnerstag, 2. August 2007 12:55 schrieb Paul Alfille:
>
> I can see the argument both ways. Part of the problem is that our
> filesystem model really matched the 1-wire design -- no inherent order to
> the entries.
>
> Do you think rewiring is likely/possible on the fly? Could we do both
> meth
On 8/2/07, Matthias Urlichs <[EMAIL PROTECTED]> wrote:
>
> Doug Collinge:
> > Presumably, the reason you are using chaining is so that you don't have
> to
> > deal with device IDs. So one natural way to interface a chain is to
> make a
> > "chain" directory and put the chained devices inside it, n
Hi,
Doug Collinge:
> Presumably, the reason you are using chaining is so that you don't have to
> deal with device IDs. So one natural way to interface a chain is to make a
> "chain" directory and put the chained devices inside it, named "0", "1",
> "2", ... This way the order of the chain is na
10 matches
Mail list logo
