1 |
Volker Armin Hemmann <volkerarmin@××××××××××.com> posted |
2 |
200901072311.14423.volkerarmin@××××××××××.com, excerpted below, on Wed, |
3 |
07 Jan 2009 23:11:14 +0100: |
4 |
|
5 |
> it is way easier to rm&untar one directoy, than to have to hunt down 213 |
6 |
> binary packages. |
7 |
|
8 |
... And you missed the vital bit, that for "live" packages the binpkgs |
9 |
get overwritten at every update. Thus, the tarball idea, either tracking |
10 |
down those 213 (using your number) binpkgs and tarballing or otherwise |
11 |
archiving them so if the update screws them there's a fallback, or doing |
12 |
the much less complex "let's just tarball the whole subtree, easy since |
13 |
it's specific to the kde version anyway", becomes the only effective |
14 |
backup for "live" packages that normally change each time they are merged. |
15 |
|
16 |
I ran into this back when I was doing the pre-4.0 livepkgs. I normally |
17 |
use binpkgs (and as I've posted numbers of times, consider them one of |
18 |
portage's best-kept secrets) for the backup potential, among other |
19 |
things, but it just doesn't work that way with "live" packages, so there, |
20 |
if the "live" versions are changing as fast as KDE's have been and often |
21 |
break as KDE's were at that point, one either has to devise and alternate |
22 |
backup method, or will inevitably run into trouble at /some/ point. |
23 |
|
24 |
That, in addition to the fact that to this point, KDE 4.x has simply been |
25 |
too broken to really do much with for those of us that tend not to be |
26 |
content with "the defaults", thus the reason KDE attracted us in the |
27 |
/first/ place, has been a big reason I decided it wasn't worth the hassle |
28 |
to keep the kde4 live versions running here, and unmerged them. |
29 |
|
30 |
>> Personally I think it would be a lot simpler for users and us to only |
31 |
>> have - kdeprefix in portage, and maintain live/snapshots that can |
32 |
>> install into kdeprefixes. I myself maintain a stable installed in /usr/ |
33 |
>> and a live tree in /usr/kde/live... |
34 |
|
35 |
That works just fine, until you try to have 3.x and 4.x installed |
36 |
together. Then 3.x in its own prefix tries to use the 4.x settings in |
37 |
the non-prefix /usr, and broken functionality ensues. There's a Gentoo |
38 |
bug about that, that I'm sure most folks trying to run both are already |
39 |
familiar with, but what I can't figure out is how upstream can live with |
40 |
it being that broken, since they're the ones that keep saying use kde3 |
41 |
stuff where kde4 is still broken -- but when kde4 is breaking kde, that's |
42 |
kind of difficult! Unless upstream has it working and something that |
43 |
Gentoo's doing breaks it? |
44 |
|
45 |
> so if you want to install 4.1 and 4.2 at the same time (because 4.2 will |
46 |
> certainly be troublesome until .2 or .3) you are screwed? |
47 |
> |
48 |
> http://marc.info/?l=gentoo-dev&m=123108542712650&w=2 |
49 |
> |
50 |
> " - decide if we should provide +kdeprefix for stable releases and |
51 |
> snapshots" |
52 |
> |
53 |
> sounds bad ... 'if' for snapshots... ? bad, bad. |
54 |
|
55 |
Well, keep in mind that what they're discussing is NOT whether the flag |
56 |
should be there, but whether it should default to enabled (+kdeprefix) |
57 |
for those who don't have the flag set either way in make.conf. Thus, in |
58 |
theory (there's that multi-hundred-package issue, but pre-set list files |
59 |
that can be simply dropped in package.use help there), it should still be |
60 |
possible for users to use package.use to set +kdeprefix as desired, thus |
61 |
enabling them to run multiple versions if desired, no matter which way |
62 |
the debate on kdeprefix default-use goes. |
63 |
|
64 |
-- |
65 |
Duncan - List replies preferred. No HTML msgs. |
66 |
"Every nonfree program has a lord, a master -- |
67 |
and if you use the program, he is your master." Richard Stallman |