From: Martin Schlemmer <azarah@gentoo.org>
To: "C. Brewer" <cbrewer@stealthaccess.net>
Cc: Gentoo-Dev <gentoo-dev@gentoo.org>
Subject: Re: [gentoo-dev] udev implementation
Date: Tue, 21 Oct 2003 22:36:19 +0200 [thread overview]
Message-ID: <1066768579.11872.163.camel@nosferatu.lan> (raw)
In-Reply-To: <200310202256.34889.cbrewer@stealthaccess.net>
[-- Attachment #1: Type: text/plain, Size: 3496 bytes --]
On Tue, 2003-10-21 at 07:56, C. Brewer wrote:
> I tried out the 0.2 version of udev today, and I realize that its way rough
> so early in the development, but I must say I was disappointed with it's
> current implementation ( and the lousy attitude of the udev FAQ "if you don't
> like it stick with devfs" didn't help). Currently I have some small concerns
> about adopting this as a whole ( somewhere on down the line)-
>
Right.
> 1)The present package consists of a tarball with just about every device node
> you could make (excepting small things like sound, ppp, more than 4 ttyS*'s)
> Is this going to be a standard, or will some form of intuitive /dev entries be
> imp'd? IIRC, the tarball is about 1.4k device nodes, and I think I need 100
> on the outside.
>
Problem is that you need sysfs support, and currently only the scsi and
major block/char devices supports it (no input, sound, etc).
The tarball is only the initial stage, when better support is there (and
I have obviously learned a lot more :), it will be dropped.
> 2) Since this won't automatically create these nodes ( unless a hotplug event
> occurs), or load the dependent modules, doesn't this seem like a step back to
> the old system, but with a name-mapping steroided hotplug?
>
Depends, creating specific entries in /sys/ will also cause these, and
when all drivers support sysfs ....
> 3) Don't get me wrong..I'm not flaming the package,and I realize devfs is crap
> as well..but the score is devfsd( crap but makes nodes and loads mods on the
> fly) and udev (maps names and supposedly does stuff with hotplugging that
> hotplug never amounted to.( and is dev'd by the hotplug peeps?ironic)). All
Eventually udev will do this as well (for me example, if with new udev,
and not having /sbin/udev in /proc/sys/kernel/hotplug, it auto loads
usb-storage + co, and creates /dev/sdc* [after deleting them of
course]). You basically just have to think back initial devfs stage =)
> that aside, what is udev going to do for the desktop? I have devices I could
> swap(USB) but with most comps coming with like 6 usb ports, I cant see more
> than some pendrive swapping at user level. Yeah, I know theres peeps out
> there with 80 pendrives and 8 hot-swappable hdd's, but is this the majority
> of users? For the likely many of us who dont need to swap and have had the
> same hardware on the same nodes that dont ever change..what does udev bring
> to the table?
>
When the driver register, it will still create /sys/ entries, and thus
the nodes you wish for (when it supports sysfs).
> Forgive me if I've gone delusional.. I was just under the impression that udev
> was going to do everything that devfsd does now _and_ add name mapping, and
> apparently I was wrong. I'm just planning for the future since seeing the
> udev changes going into our init system.. we got no choice about the devfs
> and I feel it's going the same way for udev. I'm not trying to slight the
> obviously hard work that was put into it, but what about choice? to devfs or
> not to devfs? to udev or not to udev? Or is it merely choice with package
> selection, and not with the overall package that is Gentoo?
>
If you want to do testing, and do not mind the slight issue, go udev -
if not, go devfs for now.
Thanks,
--
Martin Schlemmer
Gentoo Linux Developer, Desktop/System Team Developer
Cape Town, South Africa
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]
next prev parent reply other threads:[~2003-10-21 20:35 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-10-21 5:56 [gentoo-dev] udev implementation C. Brewer
2003-10-21 6:17 ` Jon Portnoy
2003-10-21 15:14 ` C. Brewer
2003-10-21 17:09 ` Jon Portnoy
2003-10-21 7:43 ` Christian Birchinger
2003-10-21 7:49 ` Jon Portnoy
2003-10-21 8:01 ` Ernst Herzberg
2003-10-21 20:36 ` Martin Schlemmer [this message]
2003-10-21 21:45 ` C. Brewer
2003-10-22 2:33 ` Mike Frysinger
2003-10-22 2:50 ` C. Brewer
2003-10-22 4:27 ` Martin Schlemmer
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1066768579.11872.163.camel@nosferatu.lan \
--to=azarah@gentoo.org \
--cc=cbrewer@stealthaccess.net \
--cc=gentoo-dev@gentoo.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox