thelinuxlink.net

Mayhaps my arch problem was much easier to correct than I thought. I am hearing that the new ide subsystem in the kernel, pata, is now using sd* to identify ide drives. So what was /dev/hda1 is not /dev/sda1. Anyone else hear about this?

I do remember this, though only vaguely. Wish I could remember the reason, but all I remember is that it sounded sensible at the time I read it.

dann wrote:Mayhaps my arch problem was much easier to correct than I thought. I am hearing that the new ide subsystem in the kernel, pata, is now using sd* to identify ide drives. So what was /dev/hda1 is not /dev/sda1.

That is correct, they make everything look like a SCSI device now, even standard IDE drives. I think it's to make things more consistent and coherent in that stuff use the same subsystem etc.

dann wrote:Anyone else hear about this?

Yeah, it was introduced in Fedora 7 a year ago, been using it ever since, has worked fine for the entire time. A new wireless stack was also introduced in Fedora 7, which has improved the state of wireless in Linux immensely.

Tsuroerusu wrote:

dann wrote:Anyone else hear about this?

Yeah, it was introduced in Fedora 7 a year ago, been using it ever since, has worked fine for the entire time. A new wireless stack was also introduced in Fedora 7, which has improved the state of wireless in Linux immensely.

Don't forget the new firewire stack added in Fedora 7, I have yet to see any DV grabbing tools that can work with it.

Yeah - maybe I wasn't clear enough in my post in your blog where I was referring to persistant naming. There have been many changes in the last year like this which I find absurd - I've been having a very difficult time with the "organic" nature of linux lately, to say the least. I guess I'm at a point where "nifty" is way way less important to me than well planned and stable. I've been dicking with some of the other free alternatives lately, and while I haven't been drawn enough to "switch" I'm seeing a lot that I like in both solaris and bsd, and both of them seem to be very viable desktop alternatives.

Snarkout wrote:I'm seeing a lot that I like in both solaris and bsd, and both of them seem to be very viable desktop alternatives.

Just curious, could you elaborate a little on this? I'd be interested to hear what you think about BSD specifically.

Actually, not really - I like the design of both systems, but so far have only really been tinkering with them. I like ports a lot, but still contend that it's a pita to figure out which packages are available and from where when using BSD. I like some of what PCBSD is doing, a lot. Mostly what I like is that I gather there is a great deal more concern given to backwards compat and more concern for not introducing regression. This could be BS for all I know, though - you obviously have a lot more experience with BSD while my main experience comes from admining some oldass BSDI boxen. I've been using solaris in some form or another for quite a while - my dad used to work for SUN. Now that it's "free" and now that Ian Murdoch is reporting that there will be an aptish package manager for it in the future, I'm kind of excited. Obviously the solaris HCL is microscopic when compared to linux, though. Mosthy, I just wish that the devs of the kernel and various projects I use would think more about how what they do is going to affect me as a user and admin - dealing with regressions caused by "HEY, LETS JUST CHANGE *THIS*, FUCK THE END USER" bugs the living shit out of me.

And, yes, obviously this isn't something that only happens in the linux world - hell, it happens in the Cisco world constantly, and let me tell you, it's no joke when something that worked perfectly for years fails mysteriously after an upgrade and it affects thousands of users.

dann wrote:Mayhaps my arch problem was much easier to correct than I thought. I am hearing that the new ide subsystem in the kernel, pata, is now using sd* to identify ide drives. So what was /dev/hda1 is not /dev/sda1. Anyone else hear about this?

This is a kernel thing. Apparently the SCSI subsystem is much better code than the IDE subsystem, so they're moving everything to one tree. It's been talked about for a while.

My Fedora system has been using it since 7, but Debian testing hasn't made the change yet. I imagine once Lenny goes stable they'll put it into testing.

Gomer_X wrote:

dann wrote:Mayhaps my arch problem was much easier to correct than I thought. I am hearing that the new ide subsystem in the kernel, pata, is now using sd* to identify ide drives. So what was /dev/hda1 is not /dev/sda1. Anyone else hear about this?

This is a kernel thing. Apparently the SCSI subsystem is much better code than the IDE subsystem, so they're moving everything to one tree. It's been talked about for a while.

Another argument for it could be, that using this new approach, you only have to "talk" one language, that being the SCSI protocol, when you're talking to a storage device. I am not a storage expert of any stretch of the imagination, this is just a theory.

Tsuroerusu wrote:Another argument for it could be, that using this new approach, you only have to "talk" one language, that being the SCSI protocol, when you're talking to a storage device. I am not a storage expert of any stretch of the imagination, this is just a theory.

Right. They've wanted to unify on one protocol for a while, and they picked SCSI because the code is just better.