On Mon, 2005-01-17 at 10:07 -0500, Dan Williams wrote: > No problem. It seems to be the _frequency_ of down/ups that NM did. > Since NM does them programmatically, it does them much faster and at > more random intervals than you as the user can do them. Plus, NM > programmatically sets the options on the card faster than you as the > user can type 'iwconfig eth1 xxx xxx'. This exposes race conditions in > the kernel and the drivers that were never uncovered before, since > people haven't done anything like NetworkManager before. These races, > assumptions, and errors need to be fixed or worked around. OK, beat me with a clue stick if you want, but: a) You say rapid changes of configuration OLD-NM did triggered races/bugs. b) therefor, NEW-NM doesn't do this anymore (or at least not so often). c) I have no problem at all, when I configure my prism54 card once and leave it alone. d) I had no problems at all, when OLD-NM configured my prism54 card and did whatever it wanted to do afterwards and how often it wanted to do. e) My card stops after a short time with NEW-NM. I don't see how doing up/down _less_ can break my card, when doing it _never_ works flawlessly (I could think of a situation, where OLD-NM would reconfigure the card so often, that it never had time to break in between, but then I don't see how [c] can hold true) I am usually not so dumb. jtl
Attachment:
signature.asc
Description: This is a digitally signed message part