1 |
> |
2 |
> My overall impression is that this distro really *kicks ass*. |
3 |
> |
4 |
It *does*, doesn't it? <g> Glad to hear you're enjoying it! |
5 |
|
6 |
|
7 |
> |
8 |
> Is it possible to take advantage of multitasking in Linux kernel to |
9 |
> compile and download at the same time? |
10 |
> |
11 |
Oh definately. I'm fairly certain this has been suggested before as well. |
12 |
While the idea is very good, it is the implementation that is troublesome. |
13 |
I'll defer this to drobbins to further comment on. But I would note that |
14 |
1) the bootstrap thing is essentially a script of emerge commands and |
15 |
2) emerge supports a --fetchonly option, which as you would guess, |
16 |
downloads only; no compiling. This option is popular amongst several |
17 |
of our users with dialup Internet access. |
18 |
|
19 |
|
20 |
> |
21 |
> Perl 5.6.0 <snippage> /etc/hosts |
22 |
> |
23 |
This was actually discussed on the list recently. I think it was decided |
24 |
that the install docs will be updated. I'm not certain why we're not including |
25 |
this record in our default /etc/hosts (I'd support that choice). Note to those |
26 |
interested, in /etc/hosts 'localhost' should be *after* localhost.localdomain, |
27 |
never before it! Ie: 127.0.0.1 localhost.localdomain localhost. So if we |
28 |
decide to do that, it must be added thusly, not the other way around! =) |
29 |
|
30 |
|
31 |
> |
32 |
> I've seen numerous complaints about pharmacy package. |
33 |
> |
34 |
This was corrected a couple of days ago. If you 'emerge rsync' you'll |
35 |
get the "fixed" ebuild and will be able to install it. Note: portage allows |
36 |
a package name to contain one hyphen in it's version part, and one |
37 |
optionally in the revision part. |
38 |
|
39 |
|
40 |
> |
41 |
> If the output from bootstrap/emerge is not logged, it should be! |
42 |
> |
43 |
I personally dont find that neccesary, because it's already doable by the |
44 |
end-user anyway, if they wish, via either shell redirection or something like |
45 |
nohup. Ie: either issue the command thusly: "command > output.out 2>&1 &" |
46 |
or make a shell script with the work you wanna do: |
47 |
|
48 |
-- myfile.sh -- |
49 |
#!/bin/sh |
50 |
echo "DOING bootstrap" |
51 |
bootstrap.sh bootstrap-packages |
52 |
echo "DOING system" |
53 |
emerge system |
54 |
-- myfile.sh -- |
55 |
|
56 |
and execute it like this: "nohup myfile.sh &", which logs everything to the |
57 |
file nohup.out in the current directory. I personally like this way because |
58 |
it ensures the work gets done if you log out, etc. The 'nohup' command is |
59 |
included on the buildimage.tbz2 build system, so I find it redundant to make |
60 |
portage do that unconditionally. Others may feel differently. |
61 |
|
62 |
|
63 |
> |
64 |
> some kind of total progress indicator would be helpful |
65 |
> |
66 |
Another noble idea, but not so 'easy as it sounds'. There's no decent |
67 |
candidate for 'work units' other than packages themselves. Im not sure |
68 |
how useful that really is because comparing say glibc to less is not even |
69 |
fair :P. And how do you indicate that? Ie: a simple "echo", a full blown |
70 |
ncurses interface? Another high-level question, so again, I anticipate |
71 |
others to reply, and I choose to simply tail -f nohup.out to see where I am. |
72 |
However I'm fairly certain that something like this is basically inevitable, |
73 |
ie: it will probably be implemented in one form or another eventually! =) |
74 |
|
75 |
|
76 |
> |
77 |
> nothing in /usr/src <snippage> I wonder why? |
78 |
> |
79 |
I'm farily certain the linux-headers are downloaded during bootstrap, not |
80 |
emerge system. Indeed its because glibc depends on the kernel headers. |
81 |
We (as blessed by Linus himself, and in accordance with "the right way") |
82 |
install the kernel headers used to build glibc into /usr/include/{linux,asm}, |
83 |
not in the older, libc5 tradional symlinks to /usr/src/linux way. Too many |
84 |
reasons to laundry list here, but rest assured this is the "right place" for |
85 |
kernel headers; at least at this point in time. |
86 |
|
87 |
|
88 |
> |
89 |
> "unresolved symbol" errors from depmod. |
90 |
> |
91 |
Maybe you just need to run: "depmod -a"? If not, we'll have to investigate |
92 |
a little here. Anybody else have a suggestion for this? |
93 |
|
94 |
|
95 |
> |
96 |
> I think it should be static by default. Generally, I think default configs |
97 |
> can be much better. I am talking about 2.4.10-r1 kernel. |
98 |
> |
99 |
Well if you want, just "rm .config" *before* you run make menuconfig. I'm |
100 |
not exactly sure why the default .config is the way it is, but I agree it can |
101 |
be slightly annoying to setup the kernel when you try to menuconfig with |
102 |
our current .config. Drobbins, remember I mentioned this same thing just |
103 |
the other day? Also, I've seen others hitting the 'IDE as module" thing; |
104 |
I agree we should do some tweaking in this respect. As for everything |
105 |
being static in the kernel, I personally prefer to modularize as much as I |
106 |
can, depending on what I'm building (a workstation, a fileserver, a NAT |
107 |
box, etc, etc). For one, it makes the kernel smaller, which also means |
108 |
it loads faster, it runs faster, and it uses less memory. This is a personal |
109 |
perferenece issue so YMMV. |
110 |
|
111 |
|
112 |
> |
113 |
> Emerging vim installed XFree86 4.1.0?!!! Holy cow, man! |
114 |
> |
115 |
This is probably related to you having X in your USE variable. This is |
116 |
how we control the build-time support of various goodies in Gentoo Linux. |
117 |
This is also how we can do cool stuff like compiling/installing a truckload |
118 |
of stuff with one simple command, tailored to your specific tastes. Example, |
119 |
take a fresh system, setup your USE variable, and emerge a high-level |
120 |
package, like "emerge kde-base/kdebase". That'll get a *lot* accomplished |
121 |
with a few quick keystrokes =) |
122 |
|
123 |
|
124 |
> |
125 |
> There is a mistake in your installation doc. In Code Listing 21 you say: |
126 |
> # mv bzImage bzImage.orig but you probably mean |
127 |
> # mv /boot/bzImage /boot/bzImage.orig |
128 |
> |
129 |
OK, I've fixed this on cvs. |
130 |
|
131 |
|
132 |
Thanks a lot for your feedback and suggestions. =) |
133 |
|
134 |
|
135 |
Cheers |
136 |
-- |
137 |
Donny |