1 |
On 08/09/2011 09:15 PM, Donnie Berkholz wrote: |
2 |
> On 19:42 Tue 09 Aug , Fabian Groffen wrote: |
3 |
>> On 09-08-2011 12:32:57 -0500, Donnie Berkholz wrote: |
4 |
>>>>> Yeah, that's already on my draft agenda [1]. =) But we should still have |
5 |
>>>>> a small set of options to choose from if we do vote to automate, so we |
6 |
>>>>> don't sit around for another month or discuss it aimlessly for hours. |
7 |
>>>>> Being prepared is what I'm hoping we can do here. |
8 |
>>>> |
9 |
>>>> Ok, then I suggest simply adding ", don't bother about changes between |
10 |
>>>> CVS log and ChangeLog" to both of your options. |
11 |
>>> |
12 |
>>> I guess I don't understand something here. If we aren't retroactively |
13 |
>>> changing existing ChangeLogs, and we're autogenerating things in the |
14 |
>>> future, where would these changes come from? |
15 |
>> |
16 |
>> so you want to retain all existing ChangeLogs? |
17 |
> |
18 |
> Seems like a better idea to me, although it's not originally mine. Old |
19 |
> commit messages weren't written with the knowledge or intent that anyone |
20 |
> would be reading them, except maybe a dev or two, so we might lose a lot |
21 |
> of information. |
22 |
|
23 |
Quite the opposite, as commit messages have always been targeted for |
24 |
developers as notes |
25 |
Where as ChangeLog has been NEWS to users |
26 |
So I would assume commit messages contain the more important information |
27 |
to keep the package maintaince going |
28 |
|
29 |
> |
30 |
> If/when we switch to git, we might want to reconsider that, since all |
31 |
> the handwritten messages will be old, largely irrelevant history by |
32 |
> then. |
33 |
> |