is magicdev evil?
Christopher A. Williams
chrisw01 at privatei.com
Mon Oct 6 21:27:01 UTC 2003
On Mon, 2003-10-06 at 15:14, Owen Taylor wrote:
> On Mon, 2003-10-06 at 16:57, Tommy McNeely wrote:
<snip...>
> > Don't CDROMS send some sort of "signal" when they are opened/closed that
> > would be better than constantly probing it?
>
> Basically:
>
> A) If magicdev is eating 25% of your CPU, something is going wrong,
> and you shoudl file a bug so we can fix it. It isn't supposed
> to do that.
<snip...>
>
> B) No, CDROMS do not reliably send a signal when they are opened
> and closed. There are things in the spec for that, they are not
> implemented uniformly.
>
> There may be better things to do than what magicdev does. They
> probably would require full-time work for about a month for a
> pair of a test engineer and a developer to get going and make
> reliable.
<snip...>
Of note, I have run Windows (NT/XP/Server) in a VMware session for
years. If you follow what's happening on VMware, you learn quickly that
polling a CD-R/RW/DVD device every few seconds is the norm there as
well. This shouldn't come as a surprise at all.
I noticed that magicdev seems to really interfere with XCDRoast 0.98a14
but not the earlier version. GNOME Toaster is also OK. I noticed on the
XCDRoast page that 0.98a14 apparently includes a totally rewritten
subsystem for disk reads and writes. Part of the problem could be
there...
Cheers,
Chris
--
====================================
"If you get to thinkin' you're a person of some influence,
try orderin' someone else's dog around."
--Cowboy Wisdom
More information about the fedora-test-list
mailing list