On Wed, Feb 5, 2020 at 11:58 PM Kevin Kofler <kevin.kofler@chello.at> wrote:
Kamil Paral wrote:
> I have already responded to your exaggerated numbers once, and you didn't
> even reply. "Hours of difference" for "a few percent increase", let's say
> 3 hours for 3 percent increase, means 100 hours total download time.
> That's over 4 days of non-stop download. I don't consider that plausible.

I used to download Red Hat Linux (FTP edition) in > 1 week with what was
then considered a "broadband" connection. (And these days, Fedora is so much
larger than RHL used to be that even a significantly faster connection than
that will take a week to download it.)

We all used to have dial-ups, sure. I didn't use mine to download Linux OSes, but I very much remember waiting 15 minutes to get a single mp3 file. Alright, so let's conclude there used to be at least one person with this use case. The question is how many such people use Fedora right now? Can we even figure out their number somehow? And since we can't satisfy everybody, do we want to make our decisions based on this part of the audience?

You keep repeating how a few percent size change is the night and day difference for some people (it isn't, by definition it is a few percent change - if you use huge absolute numbers, it's just because your baseline is even bigger, like 100 hours vs 103 hours of download time - still a few percent change). You also complain how bloated everything is. Yet you're one of the few people caring about the KDE spin, where major applications are duplicated or triplicated.There are 3 different web browsers(!), 2 different package managers, 2 file managers. Just pruning the apps list would make a bigger difference than any compression algorithm can. It would also make sense to create a specific spin that is targeted at near-zero-bandwidth group, containing just the bare-bone system essentials and letting them install just the stuff they need, saving on bandwidth and time. If we want to care about these users, I believe this is a much better strategy, with much better gains for them, than discussing a few percent change in compression type.