1 |
On 6/12/20 4:16 AM, Luca Barbato wrote: |
2 |
> On 12/06/2020 11:04, Georgy Yakovlev wrote: |
3 |
>> +# cargo_src_configure --no-default-features |
4 |
> |
5 |
> Shall we default in not-defaulting so we can spare some boilerplate? |
6 |
I don't think so. Let me explain. |
7 |
|
8 |
First of all, this will force to explicitly micro-manage all the |
9 |
features for all the packages in the tree. |
10 |
|
11 |
Like 90% of crates I've seen ship working choice and there's no benefit |
12 |
of micro-management, and no benefit of enabling non-default features at |
13 |
all. Not all features need to be exposed as use-flags, and this was |
14 |
quite consistent situation so far. |
15 |
|
16 |
I know it may sound a bit strange but automagic in this case works with |
17 |
no downsides, unlike other languages/frameworks/build-systems. |
18 |
|
19 |
Also it will break all the tree packages which do not manage features |
20 |
directly if merged as is. |
21 |
|
22 |
Not sure if --features=default will activate default set and how it |
23 |
will react to being passed along. |
24 |
|
25 |
--no-default-features --features default |
26 |
IDK, looks kinda unnatural. |
27 |
|
28 |
I have a feeling that we'll get more boilerplate if we pass |
29 |
--no-default-features than if we don't. |
30 |
|
31 |
We can re-evaluate as time goes by, but for now I see no major benefit |
32 |
and only downsides. |
33 |
> |
34 |
> lu |
35 |
> |