1 |
Luke-Jr wrote: |
2 |
> -----BEGIN PGP SIGNED MESSAGE----- |
3 |
> Hash: SHA1 |
4 |
> |
5 |
> On Tuesday 09 March 2004 08:36 pm, Jason Rhinelander wrote: |
6 |
> |
7 |
>>Torsten Veller wrote: |
8 |
>> |
9 |
>>>right now a -cvs ebuild (cvs.eclass) breaks if the computer is offline |
10 |
>>>and ECVS_SERVER!=offline. (Today it happens during a revdep-rebuild.) |
11 |
>> |
12 |
>>How is that broken behaviour? If a non-cvs ebuild can't download its |
13 |
>>sources, it fatals as well. |
14 |
> |
15 |
> revdep-rebuild, I would hope, only rebuilds the currently installed |
16 |
> application. With non-cvs ebuilds, the user will likely already have the |
17 |
> distfiles needed. A cvs one shouldn't update in this case... |
18 |
|
19 |
You are assuming that "what is in distfiles/cvs-src" == "what I last |
20 |
emerged" which with cvs ebuilds can be far from the truth. |
21 |
mozilla-1.5.tar.bz2 is (hopefully) always 1.5 - I can take any 1.5, |
22 |
build it, and get the same thing. "distfiles/cvs-src/mozilla" isn't. |
23 |
|
24 |
You're also wrong about how revdep-rebuild works - it doesn't attempt to |
25 |
build exactly the same thing you installed before, it merely attempts to |
26 |
rebuild any packages that seem to have broken .so's or binaries. I |
27 |
could have radically changed my use flags, and a revdep-rebuild will |
28 |
emerge the package with my new use flags, not the old ones. As it should. |
29 |
|
30 |
>>>Could you change this to something like: |
31 |
>>>If cvs-server is not available or you are not online, print a |
32 |
>>>warning/hint and try to build from an old checkout. |
33 |
>> |
34 |
>>If emerge is unable to download updated cvs sources, it seems to me that |
35 |
>>it should be handled the same way non-cvs ebuilds handle not being able |
36 |
>>to download their package sources. What you seem to be suggesting would |
37 |
>>be analogous to the mozilla-1.6 ebuild using mozilla-source-1.5.tar.bz2 |
38 |
>>because it couldn't download mozilla-source-1.6.tar.bz2. |
39 |
> |
40 |
> Rather, the way it is now would be analogous to revdep-rebuild trying to pull |
41 |
> in mozilla-1.6 when currently only mozilla-1.5 is installed. |
42 |
|
43 |
Blindly trusting that what is in cvs-src is 1.5 could lead to this exact |
44 |
effect. |
45 |
|
46 |
|
47 |
Anyway, let's get back to the original point for a second: |
48 |
|
49 |
>right now a -cvs ebuild (cvs.eclass) breaks if the computer is offline |
50 |
>and ECVS_SERVER!=offline. (Today it happens during a revdep-rebuild.) |
51 |
|
52 |
Isn't that really your own fault? How is cvs.eclass supposed to know |
53 |
your connection is down as opposed to the CVS server's? If I emerge |
54 |
mozilla-cvs, but the mozilla CVS server was down and the latest sources |
55 |
I have are a year old, should it just continue away and build a year-old |
56 |
Mozilla? |
57 |
|
58 |
|
59 |
-- Jason Rhinelander |
60 |
-- Gossamer Threads, Inc. |
61 |
|
62 |
-- |
63 |
gentoo-dev@g.o mailing list |