| 1 |
On 8/23/21 11:05 AM, Rich Freeman wrote: |
| 2 |
> On Mon, Aug 23, 2021 at 10:36 AM Ulrich Mueller <ulm@g.o> wrote: |
| 3 |
>> |
| 4 |
>>>>>>> On Mon, 23 Aug 2021, Anthony G Basile wrote: |
| 5 |
>> |
| 6 |
>>>>> **WARNING** |
| 7 |
>>>>> |
| 8 |
>>>>> If you happen to have an INSTALL_MASK with a blanket "*systemd*" |
| 9 |
>>>>> glob, you will inevitably break your system. sys-fs/udev contains |
| 10 |
>>>>> "systemd" in some of its filenames, hence a blanket filter rule will |
| 11 |
>>>>> likely lead to a non-functional udev installation. |
| 12 |
>>>> |
| 13 |
>>>> Will an INSTALL_MASK of "/usr/lib/systemd /etc/systemd" cause any |
| 14 |
>>>> issues? |
| 15 |
>> |
| 16 |
>>> I have not tested, but I think so since "systemd-" is used as a prefix |
| 17 |
>>> for files installed by sys-fs/udev. |
| 18 |
>> |
| 19 |
>> So, we've abandoned the systemd USE flag, and I remember that one of |
| 20 |
>> the arguments was that users could use INSTALL_MASK for precisely the |
| 21 |
>> above mentioned directories. |
| 22 |
> |
| 23 |
> Well, the argument is that we don't use USE flags to prevent packages |
| 24 |
> from installing small text files. It is the same reason we don't have |
| 25 |
> an openrc USE flag to control installing init.d scripts. We're now |
| 26 |
> talking about pretty far back in history but I think this was a |
| 27 |
> general guideline before systemd even came along. |
| 28 |
> |
| 29 |
>> Now the message is that users' systems will be broken if they had |
| 30 |
>> followed our previous advice? Seriously? |
| 31 |
> |
| 32 |
> Did we ever officially advise people to use INSTALL_MASK at all? I |
| 33 |
> thought that was mostly a "you can keep the pieces if you break |
| 34 |
> things" option we provide. IMO the risks of people misusing it are |
| 35 |
> far greater than the possible harm of having a few hundred small text |
| 36 |
> files installed on their system, but it is there if people really want |
| 37 |
> to use it. |
| 38 |
|
| 39 |
I remember this discussion well. It was for those "stubborn" people who |
| 40 |
wanted a clean system. I added to the discussion by saying "what about |
| 41 |
embedded systems people where every file counts because of inode and |
| 42 |
block allocation constraints" and the answer was INSTALL_MASK, not a USE |
| 43 |
flag, for the reasons Rich stated. This was to create a openrc/systemd |
| 44 |
agnostic system. |
| 45 |
|
| 46 |
Having said that, I'm open to whatever solution/wording you might suggest. |
| 47 |
|
| 48 |
> |
| 49 |
> However, having used the option in the past shouldn't hurt anybody. |
| 50 |
> It only impacts people if they use it when they install udev, hence |
| 51 |
> the news item. |
| 52 |
> |
| 53 |
|
| 54 |
|
| 55 |
-- |
| 56 |
Anthony G. Basile, Ph.D. |
| 57 |
Gentoo Linux Developer [Hardened] |
| 58 |
E-Mail : blueness@g.o |
| 59 |
GnuPG FP : 1FED FAD9 D82C 52A5 3BAB DC79 9384 FA6E F52D 4BBA |
| 60 |
GnuPG ID : F52D4BBA |
Archives