nomail-nospam@no.org wrote:
> The hwprobe=-modules.pata kernel argument workaround in the suse-10.3
> release notes is one way to go that I could live with but I'd like to be
> able to count on this workaround not getting written out of future
> versions and being available in other distros as well, at least in
> Slack, Suse & Debian (if necessary since Deb-4.0r3 & Slack-12.0 still
> use Hdx devs but who knows for how long).

This has to do with udev, just edit the configuration, you could make
the sd* from libata to appear as hd*. There will go a good while before
udev will be replaced in the same way as devfs has been.


> My 2 cents is that this should never have happened in the first place
> and I'd vote for it being fixed before it too becomes a huge immovable
> legacy issue like the number of primaries. Progress is about increasing
> or elliminating limits, not digging them up from jurassica.

Myself I think it's nice to get rid of ide-scsi stuff to get a CDrw to
work properly and even with older versions of different applications, so
the kernel developers has fixed something. :)


> What I would propose is that distros provisionally commit to keeping
> workarounds and that at least a long-term objective of upgrading libata
> be decided upon if only for the sake of over-the-horizon steering.

It's difficult to say which distro will keep on having a million
different ways to name a device, they will drop legacy methods as soon
as the majority of applications will use other methods, but you can
always edit your /etc/udev stuff.


--

//Aho

J.O. Aho wrote:
> nomail-nospam@no.org wrote:
>> The hwprobe=-modules.pata kernel argument workaround in the suse-10.3
>> release notes is one way to go that I could live with but I'd like to
>> be able to count on this workaround not getting written out of future
>> versions and being available in other distros as well, at least in
>> Slack, Suse & Debian (if necessary since Deb-4.0r3 & Slack-12.0 still
>> use Hdx devs but who knows for how long).
>
> This has to do with udev, just edit the configuration, you could make
> the sd* from libata to appear as hd*. There will go a good while before
> udev will be replaced in the same way as devfs has been.

This will change the names but won't give more partitions I
think.

>> My 2 cents is that this should never have happened in the first place
>> and I'd vote for it being fixed before it too becomes a huge immovable
>> legacy issue like the number of primaries. Progress is about
>> increasing or elliminating limits, not digging them up from jurassica.
>
> Myself I think it's nice to get rid of ide-scsi stuff to get a CDrw to
> work properly and even with older versions of different applications, so
> the kernel developers has fixed something. :)
>
>
>> What I would propose is that distros provisionally commit to keeping
>> workarounds and that at least a long-term objective of upgrading
>> libata be decided upon if only for the sake of over-the-horizon steering.
>
> It's difficult to say which distro will keep on having a million
> different ways to name a device, they will drop legacy methods as soon
> as the majority of applications will use other methods, but you can
> always edit your /etc/udev stuff.

The different names Sdx/Hdx are a minor problem for example
when trying to share/migrate scripts between different
distros. The real stinker is the lack of 64 partitions that
were there before, that the sata drives can support (and
then some), but which the libata replacement doesn't. To me
that's like buying a four-door car and then welding two of
them shut :-)