1 |
>>>>> On Wed, 18 Jul 2018, Christopher Head wrote: |
2 |
|
3 |
> On July 18, 2018 2:55:55 AM PDT, Ulrich Mueller <ulm@g.o> wrote: |
4 |
>> It was mentioned that all three directories (ebuild repository, |
5 |
>> binary packages, distfiles) have some characteristics of a cache. |
6 |
>> However, I think this is much more true for distfiles than for the |
7 |
>> other two, which cannot be easily restored to a previous state. |
8 |
|
9 |
> Neither can a Web browser’s cache, if the remote page has changed, |
10 |
> yet those are still called caches. Surely a cache is something that |
11 |
> can safely be discarded without negative impact |
12 |
|
13 |
In my understanding, a cache is typically an open collection of items. |
14 |
Some subset of them can be deleted without much negative consequence, |
15 |
and there may also be surplus items that are no longer necessary and |
16 |
will be expired at some later time in order to reclaim disk space. |
17 |
|
18 |
Nothing of this is true for an ebuild repository, which is a closed |
19 |
collection of files: A single file cannot be discarded without |
20 |
invalidating the whole repository. Also there cannot be any stray |
21 |
files which would be expired later. Same as above, a single stray file |
22 |
will invalidate all. |
23 |
|
24 |
(A collection of binary packages may qualify as a cache though, by |
25 |
this definition.) |
26 |
|
27 |
> (which, for most users, the ebuild tree can be, since for most |
28 |
> users, getting back today’s tree rather than last week’s is not a |
29 |
> negative impact), not something that can be restored to exactly its |
30 |
> prior state automatically. |
31 |
|
32 |
Consider a production system that is only updated at certain well |
33 |
defined times. Its system administrator may still want to install one |
34 |
additional package, or flip a use flag for another. In this scenario, |
35 |
the gentoo repository is a precious resource, and restoring it to a |
36 |
different state would not be acceptable. |
37 |
|
38 |
Ulrich |