1 |
On Sat, 8 Sep 2012 10:47:47 -0400 |
2 |
Philip Webb <purslow@××××××××.net> wrote: |
3 |
|
4 |
> My new machine is working & compiles lightning-fast : |
5 |
> it feels like the driver of a steam engine hurtling down the tracks |
6 |
> (I've just finished a biography of J G Robinson, the UK loco |
7 |
> designer). |
8 |
> |
9 |
> I've just recompiled most of the pkgs listed by 'emerge -ep system', |
10 |
> but found a few problems : |
11 |
> |
12 |
> (1) Gcc 4.5.4 seems to require USE="cxx", not the previous "-nocxx", |
13 |
> which was covered by "-*" at the beginning of my list in make.conf . |
14 |
|
15 |
USE=-*" is really not a good idea. It undoes all the sane defaults that |
16 |
cascading profiles give you. |
17 |
|
18 |
It is far far better to select the closest most appropriate profile |
19 |
and add any desired flags that are not in the profile. |
20 |
|
21 |
All you have succeeded in doing is removing the sane defaults and |
22 |
forcing yourself to do it all manually. Which is really a whole lot of |
23 |
heavy lifting Gentoo is designed to not require you to do. It's |
24 |
Gentoo, not LFS :-) |
25 |
|
26 |
|
27 |
> 4.5.4 compiled without "cxx" wouldn't recompile itself with "cxx" (!), |
28 |
> but I found I could get around the problem by emerging 4.4.5 , |
29 |
> then using that to recompile 4.5.4 , after which 4.5.4 did recompile |
30 |
> itself. It doesn't hurt to have > 1 version of Gcc installed. |
31 |
|
32 |
A sane profile would have prevented this from happening at all. |
33 |
|
34 |
> |
35 |
> (2) Libxml2 failed : I simply left it till tomorrow to find out why. |
36 |
> |
37 |
> (3) Groff + Openssh have an "X" flag : is this useful ? |
38 |
|
39 |
for groff this builds gxditview, whatever that is. Probably an X |
40 |
man-page viewer. I've never used it, I always run man in a terminal, I |
41 |
suspect most other folk do too. |
42 |
|
43 |
For openssh, the only thing this does is add xauth as a dependency. |
44 |
For a workstation running X, you already have xauth. |
45 |
|
46 |
So you can safely leave X enabled for both packages. |
47 |
|
48 |
|
49 |
> Python has a "build" flag, which suppressed many deps, |
50 |
> but now I'm back in my regular machine |
51 |
> 'euses' issues a grim warning vs setting it. |
52 |
|
53 |
Yes, you do not want to set USE="build" for python. You are not |
54 |
building stage 1 images |
55 |
|
56 |
> |
57 |
> (4) The ancient distinction between 'world' / 'system' |
58 |
> seems to have degenerated into chaos : |
59 |
> some pkgs which clearly sb part of 'system' aren't, |
60 |
> while others -- eg virtuals -- have been added. |
61 |
|
62 |
This was discussed to death about a year ago. |
63 |
|
64 |
Specific pagers and editors are not in system anymore. Virtuals are. |
65 |
|
66 |
The standard stage 3s give you a default pager and editor, so that if |
67 |
you remove it, portage will add it right back in (and thereby not leave |
68 |
you stranded up shit creek with no paddle in sight). |
69 |
|
70 |
If you don't like nano as you editor and like me can't abide it's |
71 |
existence, then merge vi and unmerge nano. The virtual is still |
72 |
satisfied. |
73 |
|
74 |
All of this is a good thing, it just takes a little thinking and |
75 |
acceptance on your part. |
76 |
|
77 |
> |
78 |
> Incidentally, I've found out why the system creates many TTYs : |
79 |
> they're the equivalent of GUI workspaces = desktops, |
80 |
> allowing someone working without X to view different files etc. |
81 |
> I'm continually struck by the genius of those who created UNIX in |
82 |
> 1969 ... |
83 |
> |
84 |
|
85 |
They are there because Unix was from the very start designed to be a |
86 |
multi-user, multi-tasking operating system and each user needs a tty. In |
87 |
the 60s it was unthinkable to do it any other way. I don't think this |
88 |
was any special genius on Thompson's part, it was the norm back then |
89 |
|
90 |
|
91 |
|
92 |
-- |
93 |
Alan McKinnon |
94 |
alan.mckinnon@×××××.com |