1 |
On Sun, Oct 27, 2019 at 06:58:00PM +0100, Michał Górny wrote: |
2 |
> On Sun, 2019-10-27 at 12:40 -0500, William Hubbs wrote: |
3 |
> > Most upstreams and build systems do not make this distinction, so this |
4 |
> > causes unnecessary hacks in ebuilds. |
5 |
> > |
6 |
> |
7 |
> The hacks aren't 'unnecessary'. There is a very good reason that files |
8 |
> that are used *purely at build time* don't land in /. That reason is |
9 |
> disk space. Even if people nowadays are forced to use initramfs with |
10 |
> separate /usr, it doesn't mean you should just let their rootfs fill up |
11 |
> with useless files. |
12 |
|
13 |
The useless files argument really holds no water with me. We install |
14 |
many files on / that are useless in one situation or another. |
15 |
Some examples are logrotate files when logrotate isn't installed, |
16 |
systemd units for openrc systems and openrc init scripts for systemd |
17 |
systems. |
18 |
|
19 |
Talk to me about useless files on / after we put all of these, and |
20 |
possibly others I can't think of, behind use flags. |
21 |
|
22 |
> Do you have any *real* argument? Because 'unnecessary hack' is |
23 |
> basically your feeling of ebuild aesthetics. My aesthetics is more |
24 |
> worried about useless clutter in /lib*. FHS agrees with me, as you |
25 |
> yourself admitted yesterday. |
26 |
|
27 |
Any downstream hack means that we are being lazy and not reporting the |
28 |
bug upstream and asking them to fix it. |
29 |
|
30 |
This particular issue is not a big deal to any other distro and has |
31 |
never been. Shouldn't we try to get upstreams to do this if it is so |
32 |
important? |
33 |
|
34 |
> So why do you believe we should introduce this regression? And why are |
35 |
> you trying to sneak it past most of the developers via gentoo-portage- |
36 |
> dev instead of gentoo-dev? |
37 |
|
38 |
This attack is un called for. This list is as open as any other, |
39 |
and there is no need for you to make this out to be some kind of |
40 |
conspiracy theory to "sneak" something past the developers. |
41 |
|
42 |
William |