Hi,
On 8/20/24 18:09, Bastian Venthur wrote:
That's what I thought too: we should somehow incorporate the popcon
value.
I disagree -- the users most affected by a removal are those who
automate installation, and those will not be running popcon.
Can you elaborate on that claim? I probably miss some context here, but
why do you think users most affected are the ones automating
installations and why do you think they won't be running popcon?
Desktop installations (that are most likely to be running popcon,
because the user was asked to enable it during installation) will
usually not notice a package being gone -- they no longer get updates
for it, and if they use aptitude, it moves to the "Obsolete Packages"
section.
The people who will notice first are people building Docker images,
because a missing package means that their script fails. Docker images
typically don't have popcon, because they don't go through d-i.
Popcon was introduced to determine which packages should go on the
first installation CDs, but it cannot ever produce an accurate picture
of which packages are actually used (and that was communicated clearly
back then, but seems to have been lost to the sands of time).
I think popcon does give a good approximation of how much percent of
people installed a certain package even if not everyone uses it, don't
you agree?
No. It is desktop heavy, and counts installations instead of users. This
is a feature for determining which CD image a package should go to,
because desktop users are most likely to install from CD/USB, and popcon
directly impacts the installation experience here, but it is a bad
metric for determining how many users are affected by a change.
Simon
