1 |
Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
2 |
|
3 |
|
4 |
> On Wednesday 21 July 2010 23:14:35 covici@××××××××××.com wrote: |
5 |
> > > This is a painful process. It's enough to drive a sysadmin to drink or |
6 |
> > > (god forbid), to Windows. Portage can't help as the ebuild doesn't know |
7 |
> > > what you have installed. So you must run a script to go and dig out all |
8 |
> > > this crap for you. |
9 |
> > > |
10 |
> > > |
11 |
> > > |
12 |
> > > All I can say is, every day I get down on my knees and offer thanks that |
13 |
> > > perl is not slotted. |
14 |
> > |
15 |
> > But portage should be sensible enough to either run this for you, or |
16 |
> > stop emerging -- I had a lot of trouble during the last update where I |
17 |
> > kept getting errors and I emerged a couple of them before I knew I had |
18 |
> > to run perl-cleaner. |
19 |
> |
20 |
> You haven't thought this through and haven't consider how portage knows what |
21 |
> to do. |
22 |
> |
23 |
> Portage doesn't do it because portage can't. |
24 |
> You want portage to do it != portage can do it. |
25 |
> |
26 |
> Consider this: |
27 |
> |
28 |
> [I] dev-lang/perl |
29 |
> Installed versions: 5.12.1-r1(23:11:24 21/07/10)(berkdb gdbm -build - |
30 |
> debug -doc -ithreads) |
31 |
> |
32 |
> [I] dev-perl/DateManip |
33 |
> Installed versions: 5.56(19:39:11 17/07/10)(-test) |
34 |
> |
35 |
> |
36 |
> When I upgraded perl to 5.12.1-r1, DateManip was not upgraded. Why not? |
37 |
> because it's version number did not change and that is the ONLY thing portage |
38 |
> considers. DateManip depends on perl, not on =perl-whatever-I-used-to-have |
39 |
> |
40 |
> So portage does not know of the link between these two things and cannot take |
41 |
> them into account. Portage won't be expanded anytime soon either - you saw how |
42 |
> long it took for perl-cleaner to run, must portage go through something like |
43 |
> that with every emerge? |
44 |
> |
45 |
> Similarly, one could say portage should detect rev-dep breakage. Surprise! It |
46 |
> doesn't. revdep-rebuild does that (comparable to perl-cleaner) and you know |
47 |
> how long that takes to run. |
48 |
> |
49 |
> So you wasted some time with an upgrade. Well that's a shame. But we don't |
50 |
> care much, especially if you don't read the elog messages. If you feel that |
51 |
> portage should does this automagically, and have a plan to make it run REAL |
52 |
> quick, and have proven, workable, debugged, solid, stable patches, then I'm |
53 |
> sure Zac would be very happy indeed to hear from you. |
54 |
> |
55 |
> In the meantime, read the elog messages. |
56 |
But I could not read the elog messages, the emerge was going on, till I |
57 |
got the first error and I didn't realize that portage had upgraded perl |
58 |
-- the only thing I would like portage to do is to know that something |
59 |
must be run and stop so I can do this. You could have a list of |
60 |
packages which require a stop after emerging or something. I am |
61 |
thinking out loud here, but this is what I am trying to say. |
62 |
|
63 |
-- |
64 |
Your life is like a penny. You're going to lose it. The question is: |
65 |
How do |
66 |
you spend it? |
67 |
|
68 |
John Covici |
69 |
covici@××××××××××.com |