1 |
On Mon, 2010-01-18 at 18:23 -0600, Dale wrote: |
2 |
> Iain Buchanan wrote: |
3 |
> > On Mon, 2010-01-18 at 23:25 +0200, Alan McKinnon wrote: |
4 |
> > |
5 |
> >> On Monday 18 January 2010 22:47:05 Dale wrote: |
6 |
> >> |
7 |
> >>> In that case, ctrl alt F1 does nothing. You also need to understand |
8 |
> >>> that most people don't even know how to use SysRq keys. I didn't and |
9 |
> >>> had to do a hard shutdown. I had to actually pull the plug to do any |
10 |
> >>> good. Luckily I knew how to get it to boot into single user mode so I |
11 |
> >>> could disable hal otherwise I would be right back on the same screen |
12 |
> >>> again with no mouse or keyboard. It would be really bad if even that |
13 |
> >>> didn't work with devicekit. I'm not sure how it couldn't but we never |
14 |
> >>> know do we? |
15 |
> >>> |
16 |
> >> Dale's experiences highlight a very important and very fundamental rule of |
17 |
> >> desktop system design: |
18 |
> >> |
19 |
> >> As a developer you must completely and totally guarantee to the full limit of |
20 |
> >> what is feasible, that the user will always have a usable keyboard, mouse and |
21 |
> >> display after the desktop has launched. You can fallback to VGA resolution and |
22 |
> >> the most basic keyboard layout possible if you need to, but you must give the |
23 |
> >> user something and never leave them stranded. Anything else is just an epic |
24 |
> >> fail. |
25 |
> >> |
26 |
> > |
27 |
> > My 2c worth is this: In any other distribution, the xorg/hal update |
28 |
> > would have been configured so that Dale's (sorry to keep using you as an |
29 |
> > example :) keyboard / mouse was working. But this is Gentoo. You ARE |
30 |
> > the distributor AND the end user. Conflicts in libraries / packages are |
31 |
> > up to you to resolve. |
32 |
> > |
33 |
> > About 3-4 people use Gentoo at work, and at least 2 were hit by the |
34 |
> > keyboard/mouse not working bug in xorg when it moved to HAL. With a bit |
35 |
> > of fuddling, remerging, and so on, we got it working in both cases. |
36 |
> > |
37 |
> > So yes, the developer must give a fallback method of using the |
38 |
> > keyboard / mouse, but not against the incorrectly packaged / configured |
39 |
> > system. In Gentoo you often end up with an incorrect system, hence |
40 |
> > revdep-rebuild and so on. |
41 |
> > |
42 |
> > |
43 |
> |
44 |
> I didn't distribute hal, |
45 |
|
46 |
well, in a sense you've distributed it to yourself, as opposed to using |
47 |
a binary distribution where all these packages are rebuilt by someone |
48 |
else and distributed to you. |
49 |
|
50 |
> heck, I didn't even want it really. It's |
51 |
> required by KDE is the only reason I have it at all. I just had to |
52 |
> disable it for xorg is all to get a working X. |
53 |
> |
54 |
> Surely this wasn't my fault? |
55 |
|
56 |
no, but my point was a binary OS would re-compile everything multiple |
57 |
times on some super-server of theirs before you download and try it. |
58 |
Hence in that case you're the user, not the distributor. In Gentoo's |
59 |
case you're the user AND the distributor, and 99.9% of the time you |
60 |
don't need to recompile the universe to end up with a working system. |
61 |
I'm sure that there is some magic package that just needs to be |
62 |
re-merged that would fix the issue for you, but I'm sure you've spent |
63 |
enough time on it, so I'm not suggesting you try :) |
64 |
|
65 |
-- |
66 |
Iain Buchanan <iaindb at netspace dot net dot au> |
67 |
|
68 |
The whole intent of Perl 5's module system was to encourage the growth |
69 |
of Perl culture rather than the Perl core. |
70 |
-- Larry Wall in <199705101952.MAA00756@××××.org> |