1 |
On 9/27/21 2:15 PM, Mike Gilbert wrote: |
2 |
> On Mon, Sep 27, 2021 at 1:44 PM Robin H. Johnson <robbat2@g.o> wrote: |
3 |
>> |
4 |
>> On Mon, Sep 27, 2021 at 01:15:10PM -0400, Mike Gilbert wrote: |
5 |
>>> On Mon, Sep 27, 2021 at 12:23 PM Mike Pagano <mpagano@g.o> wrote: |
6 |
>>>>> Adding linux-info calls to pkg_pretend or pkg_setup causes slowdowns |
7 |
>>>>> when running emerge, so we should do so only when there is a |
8 |
>>>>> compensating benefit. |
9 |
>>>> |
10 |
>>>> Is this a significant slowdown? Do you have any numbers? |
11 |
>>> |
12 |
>>> Adding a check for CONFIG_PACKET to the dhcpcd ebuild adds around 7 |
13 |
>>> seconds of delay time to the pkg_setup and/or pkg_pretend phase on my |
14 |
>>> system. |
15 |
>>> |
16 |
>>> That's ok if a small number of packages are doing it, but it would |
17 |
>>> become quite annoying if a significant number of them get queued up. |
18 |
>> 7 seconds is ridiculous. |
19 |
> |
20 |
> I think the horribly slow performance is mainly due to this change in |
21 |
> how we detect the kernel version: |
22 |
> |
23 |
> https://gitweb.gentoo.org/repo/gentoo.git/commit/eclass/linux-info.eclass?id=d1ea596f034285cd044006b107a60902ea059793 |
24 |
> |
25 |
> That causes "make" to be called 4 times, and each make invocation adds |
26 |
> a couple of seconds to the get_version call. |
27 |
> |
28 |
|
29 |
Changes to these eclasses are like a prison of circular breakage for which there is no escape. |