[arch-general] `base` group replaced by mandatory `base` package - manual intervention required

[Yaro Kasear]

On 10/8/19 1:45 PM, David C. Rankin wrote:
> On 10/08/2019 01:33 PM, Eli Schwartz via arch-general wrote:
>> Because this is not about containers. There are tons of things in the
>> old base group which I don't want installed on my heavyweight X11
>> desktop which is used for media consumption.
>>
>> I don't need netct (because networkmanager is love), s-nail (unuseful in
>> practice) or vi (symlink to vim) as a baseline fact.
>>
>> I don't need cryptsetup or device-mapper if I'm not opting into an
>> encrypted filesystem, but this does not matter as I cannot get rid of
>> either one due to systemd performing shared library links to
>> libcryptsetup.so and therefore also having a depends+=('cryptsetup')
>>
>> I do not need mdadm or lvm2, because I don't use RAID or lvm, so why do
>> you think my system is unusable without it? Note: in practice, both are
>> required by libblockdev, which means if you use udisks2 you have both
>> installed anyway.
> As long at it passes the Allan test, then so be it. I do use mdadm, netctl,
> s-nail (mailx) but agree with vim as baseline. The point being no kernel? So
> now a 'base' install does not result in a running system? It seems like
> forcing the install of `base` + (a list of other packages) just to result in a
> bootable system will create more problems then it solves. At least a meta of
> 'base-legacy' would provide the same install capability. As for the argument
> advances that this was due to those looking for a container install, why not
> create a 'base-container' or 'base-minimal' and leave the traditional 'base'
> alone?
>
There are multiple kernels officially supported. Vanilla, lts, zen, hardened, etc.
So it probably was left out of base because they didn't want to force people to 
take the extra step of changing kernels. 

Yaro