1 |
On Mon, Nov 12, 2007 at 01:26:46PM +0100, Marijn Schouten (hkBst) wrote: |
2 |
> Ciaran McCreesh wrote: |
3 |
> > On Fri, 9 Nov 2007 22:40:08 +0000 |
4 |
> > Ciaran McCreesh <ciaran.mccreesh@×××××××××××××.uk> wrote: |
5 |
> >> Is the following set sufficient? Is the following set the least |
6 |
> >> restrictive correct solution? |
7 |
> > |
8 |
> > ... to explain the implications of these... |
9 |
> > |
10 |
> > Say we have packages a, b and c, and none of them have any |
11 |
> > dependencies. One valid solution to the build ordering is as follows: |
12 |
> > |
13 |
> > * Install a |
14 |
> > * Install b |
15 |
> > * Install c |
16 |
> > |
17 |
> > One of many solutions that is *not* valid is: |
18 |
> > |
19 |
> > * Start doing a, b and c in parallel. Install them as they become |
20 |
> > ready, doing only one merge at once. |
21 |
> > |
22 |
> > Another that is not valid is: |
23 |
> > |
24 |
> > * Start doing a, b and c in parallel, but don't merge them. |
25 |
> > * Merge a. |
26 |
> > * Merge b. |
27 |
> > * Merge c. |
28 |
> > |
29 |
> > One that is valid is: |
30 |
> > |
31 |
> > * Build binary packages for a, b and c in parallel. |
32 |
> > * Merge a's binary. |
33 |
> > * Merge b's binary. |
34 |
> > * Merge c's binary. |
35 |
> |
36 |
> What exactly is the difference between this valid situation and the previous |
37 |
> invalid one? |
38 |
|
39 |
The state of the environment when pkg_setup is run. In the first |
40 |
situation you can't trust it (it is racy and unpredictable among other |
41 |
things). In the second one, you can. |
42 |
|
43 |
That's the first thing that I can think of, there might be others. |
44 |
|
45 |
- ferdy |
46 |
|
47 |
-- |
48 |
Fernando J. Pereda Garcimartín |
49 |
20BB BDC3 761A 4781 E6ED ED0B 0A48 5B0C 60BD 28D4 |