| 1 |
On 01/21/2013 10:38 PM, Robin H. Johnson wrote: |
| 2 |
> CONFIG_CHECK has not been fatal for some years now, because there turned |
| 3 |
> out to be some cases where it cannot detect what the system really has |
| 4 |
> [1], or what is returned is wrong [2]. |
| 5 |
> |
| 6 |
> However, while this is has been superb in helping those corner-cases, |
| 7 |
> the fallout is that users frequently ignore the non-fatal warnings that |
| 8 |
> a configuration option is needed to run a binary later, and end up with |
| 9 |
> weird breakage. |
| 10 |
> |
| 11 |
> This patch introduces a new option, CONFIG_CHECK_FATAL, defaulting to |
| 12 |
> enabled, that explicitly causes a die if: |
| 13 |
> - CONFIG_CHECK cannot be performed successfully. |
| 14 |
> - Any CONFIG_CHECK options fail. |
| 15 |
> |
| 16 |
|
| 17 |
Technical issues aside, I'm conflicted on this. In general, I don't |
| 18 |
believe in protecting people from shooting themselves. However, I also |
| 19 |
know how easy it is to ignore elog messages. |
| 20 |
|
| 21 |
I guess this change is ok, given that I can opt-out fairly easily. Zac's |
| 22 |
workaround for binary packages makes me feel better too. |
Archives