New comment by the-maldridge on void-packages repository

https://github.com/void-linux/void-packages/pull/24357#issuecomment-676805631

Comment:
Primary advantages with noarch
  * it saves a meager amount of disk space that we're somehow still worried about in 2020 when you can buy commodity 10T disks at reasonable prices delivered to your DC of choice in around 2 days.

Primary disadvantages of noarch
  * pretty much everything else about it.

In no particular order noarch relies on kludges within xbps itself, it makes managing the repos unnecessarily complicated, it makes it shockingly difficult to construct a coherent package graph that could be used to schedule builds in a distributed fashion (along with our kludgy multilib, but that's another story).

As far as timeline to drop noarch from xbps?  As soon as it is practically feasible.  Removing that code would be on the critical path to improving the dependency solver and generally making xbps more robust.  I would go so far as to propose a bulk rebuild of all current noarch packages to make it possible to dump the support from xbps sooner.

I disagree that there is no need to touch templates.  It makes sense to remove the components that trigger this behavior to minimize confusion.  We are not intending to support it, so we should not keep the detritus in place that suggests we do.