1 |
Coming back again to your previous posting. |
2 |
|
3 |
>>>>> On Mon, 9 Nov 2009, Ciaran McCreesh wrote: |
4 |
|
5 |
> The problem lies in the exceptions. Either we word PMS so vaguely |
6 |
> that it's legal for the package manager to clobber any mtime (thus |
7 |
> defeating the point of guaranteeing preservation at all), |
8 |
|
9 |
Agreed. This is not what is wanted. |
10 |
|
11 |
> or we include long, convoluted wording describing exactly the files |
12 |
> Portage currently alters |
13 |
|
14 |
Hm, maybe this isn't as bad as it seems: |
15 |
|
16 |
,---- |
17 |
| The package manager must preserve modification times of regular files. |
18 |
| This includes files being compressed before merging. Exceptions to |
19 |
| this are: |
20 |
`---- |
21 |
|
22 |
Now we need to enumerate the exceptions: |
23 |
|
24 |
,---- |
25 |
| * files newly created by the package manager, |
26 |
`---- |
27 |
|
28 |
This will cover splitdebug, for example. (And please don't tell me |
29 |
that the wording is flawed because the PM could save a file's contents |
30 |
in some buffer, then delete the file and create it newly. This would |
31 |
be as unreasonable as the rot-13 example.) |
32 |
|
33 |
,---- |
34 |
| * binary object files being stripped of symbols. |
35 |
`---- |
36 |
|
37 |
Anything else missing from above list? |
38 |
|
39 |
> (thus preventing reasonable-looking future changes), [...] |
40 |
|
41 |
I don't get the point here. For any future change not covered by the |
42 |
list of exceptions, the PM would have to preserve mtime, in spite of |
43 |
modifying the file. Why would this prevent doing the change? |
44 |
|
45 |
Ulrich |