1 |
Thierry de Coulon posted <200602100848.26651.tcoulon@××××××××.ch>, |
2 |
excerpted below, on Fri, 10 Feb 2006 08:48:25 +0000: |
3 |
|
4 |
> I sure have to get more understanding of how to run 32bit apps on a 64bit |
5 |
> distro - I am willing to learn but you'll probably have to suffer a lot of |
6 |
> questions from me. But this does not seem to be a problem on this list :) |
7 |
|
8 |
Happily, no. It's a point of pride for many of us that Gentoo gets high |
9 |
marks for the friendliness and accessibility of its user community. =8^) |
10 |
|
11 |
> I'll think it over again, but I think I'll go for a compromise (we swiss are |
12 |
> well known for this way of doing, aren't we?): I have a working Mepis 32bit |
13 |
> (the distribution I use on my "old" main machine. SO I'll probably build a |
14 |
> 64bit Gentoo and I can switch the day I feel it works the way I like. |
15 |
|
16 |
We'll all be rooting for you! |
17 |
|
18 |
Note that I took something like three months to get my Gentoo amd64 system |
19 |
up and running, in part because I had some very specific ideas of how I |
20 |
wanted it to look and work when I was done, and as I result, had to adapt |
21 |
the installation instructions somewhat. Some of those adaptations turned |
22 |
out to be mistakes, or cause something else to break, so I'd have to go |
23 |
back and "unbreak" things, and actually started over a couple times. Of |
24 |
course, the whole time I was learning all the more, because I was breaking |
25 |
things and learning how they worked and how to fix them in the process, |
26 |
and the whole time I was keeping up on the user, amd64, and devel lists |
27 |
(the latter of which I follow in part because I like to have a heads-up on |
28 |
where Gentoo is headed), so by the time I got a working system, to the |
29 |
point where I /could/ experience some of the common mistakes of others, I |
30 |
had not only read all about them and avoided making them myself, but |
31 |
understood why they could happen and a bit about why the Gentoo devs |
32 |
hadn't or couldn't simply ensure they didn't happen in the first place! |
33 |
|
34 |
This was all possible because while I only have a single system going, I |
35 |
had a fully working Mandrake for AMD64 installation on another set of |
36 |
partitions, and would routinely have a chroot session going, in which I'd |
37 |
be setting up something in Gentoo, or trying out some strange alternative |
38 |
of my own, but it was just a chroot running in a konsole window on my |
39 |
otherwise normally operating Mandrake system. Of course, the fact that |
40 |
my system is a dual Opteron helped with that, keeping all the Mandrake |
41 |
stuff responsive even as I was emerging something on the Gentoo side for |
42 |
the Nth time, because I was too stubborn to simply follow instructions. =8^) |
43 |
|
44 |
Note that this wouldn't have worked if I'd been running a 32-bit OS, |
45 |
because the 64-bit compilations wouldn't have worked the same way in that |
46 |
case, but that I had only a single computer to work on, while you can |
47 |
simply keep running your 32-bit stuff on your current Mepis install, until |
48 |
as you said, you are comfortable with the 64-bit Gentoo install on your |
49 |
new machine. |
50 |
|
51 |
Two hints I like to give to all new users: |
52 |
|
53 |
1) Gentoo documentation is considered some of the best in the community. |
54 |
Don't be afraid to use it. In particular, the Gentoo Handbook is *NOT* |
55 |
just the install section. To make the best use of your new Gentoo system, |
56 |
you'll want to read the Working with Gentoo and Working with Portage |
57 |
sections as well (altho it's OK to skim the chapters dealing with ebuild |
58 |
details and the like, just getting a bit of how ebuilds work out of them, |
59 |
as helps in troubleshooting when one /doesn't work for you). Very often, |
60 |
it's quite apparent by the questions people ask, that they skipped that |
61 |
information, and it just makes it harder for them to manage their system, |
62 |
until the go back and read over it, or pick it up a piece at a time on the |
63 |
lists and forums. To me, that's a shame, when they could have avoided |
64 |
needless pain and mistakes, and be managing their system much more |
65 |
effectively, had they just read the documentation that's there to read. |
66 |
|
67 |
2) Before you get too far along, certainly before your first emerge |
68 |
--update --deep world and preferrably before you start merging world |
69 |
packages at all, consider adding "buildpkg" to your FEATURES line in |
70 |
make.conf. What this does is covered in the handbook, so you'll understand |
71 |
more about it after reading the sections I recommended above, but the |
72 |
handbook doesn't quite convey how useful it can be in so many words, and |
73 |
people often miss it. |
74 |
|
75 |
What FEATURES=buildpkg does is simple, really. It tells portage to |
76 |
create and store a binary package for everything it emerges. What this |
77 |
means is that if you upgrade a package and find it isn't working quite |
78 |
right, it's quite easy to use emerge -K to revert to the earlier version |
79 |
that portage created a binary package of when it merged it. Without this |
80 |
binpkg, you'd have to do the entire remerge from source thing, which on |
81 |
some packages can take awhile, all to see if it corrected the problem you |
82 |
discovered with a new version, or if the problem was elsewhere. If you |
83 |
find the old version has the problem too, again, it's a very simple |
84 |
process to remerge the now binpkged new version again. (For doing the |
85 |
same thing with already merged packages, there's the quickpkg utility. |
86 |
However, you have to know you'll need it to use it, while if portage |
87 |
automatically creates the binpkgs, they'll be there for you when you need |
88 |
them, whether you thought you might need them or not.) There are some |
89 |
other advantages to buildpkg as well, such as making it easier to recover |
90 |
if gcc or even portage quits working, because you have a binary package |
91 |
available, but these are side effects of the primary benefit, that those |
92 |
binary packages are created and available for you in the first place. |
93 |
|
94 |
The buildpkg FEATURE typically requires an additional 1-4 gigs of space to |
95 |
store packages for the entire system. A gig will usually just barely |
96 |
cover one version of each package, two gigs gives you room for several |
97 |
versions of the ones changing the fastest, four gigs gives you plenty |
98 |
of room for expansion and means you won't have to worry about cleaning out |
99 |
the oldest stale versions very often. (Newer versions of gentoolkit, a |
100 |
very useful addition to portage that most folks eventually merge, include |
101 |
eclean, a tool that helps manage the packages dir, cleaning out the stale |
102 |
versions of packages. I think eclean is only in the ~amd64 version, |
103 |
however, not the stable amd64 version, yet.) I arrived at the 1-4 gig |
104 |
figure because I have my packages dir on a separate partition, here. It's |
105 |
now 4 gig for expansion, but 2 gig was usable, while a gig was simply |
106 |
cramped. |
107 |
|
108 |
-- |
109 |
Duncan - List replies preferred. No HTML msgs. |
110 |
"Every nonfree program has a lord, a master -- |
111 |
and if you use the program, he is your master." Richard Stallman in |
112 |
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html |
113 |
|
114 |
|
115 |
-- |
116 |
gentoo-amd64@g.o mailing list |