On Tue, Sep 1, 2020 at 5:23 AM Frantisek Zatloukal <fzatlouk@xxxxxxxxxx> wrote: > > Saving 142 MBs isn't going to make a huge difference in download times (disclaimer: I don't know how is the internet connection speed in other areas, I have not that fast connection of 200mbps down (slowest possible in my area), so 142 MB saving would make roughly 6 seconds difference myself). > Just for context, I also have 200Mbps down and that's not quite the fastest possible in my area but it's basically the fastest residential connection that people actually get. Most people that I know in this part of the US have 50-100Mbps and they don't always get it. The other consideration, apart from raw performance, is bandwidth caps. I won't pretend to know where places have caps, but I have heard a lot of Australian friends complain about them over the years. My understanding, at least in the US, is that providers have largely dropped the caps in These Uncertain Times[tm], but that can be a factor for people too. > On the other hand, saving minutes in the installation process (as seen with zstd if I am reading the numbers correctly) is not only going to help users with installation times, but also us, Fedora QA, with faster testing, both manual and automated (OpenQA). > This is also true. Certainly it would help *us* to have faster installation times. I'm less convinced that it's meaningful to the majority of our users. Totally guessing, but I'd suspect that most (as in 90%) Fedora users have 1-5 or mayyybe 5-10 machines and they're not frequently re-installing them. I'd love to see the distribution of machines per user, but I don't think we'll ever know that in a reliable sense. But I do think that optimizing the install time is a meaningful benefit to only a very small portion of the audience. It's an important portion, though, so we have to decide if it outweighs the benefit to those who would be better served with a smaller download. I don't think a 25 second increase in install time will be that noticeable for most people (though, yes, those seconds add up). What I think we're missing is beyond the scope of this proposal: a context for what we're trying to achieve. What targets do we have for build time, image size, and installation time? With that, we can then make decisions about how to balance tradeoffs within that context. -- Ben Cotton He / Him / His Senior Program Manager, Fedora & CentOS Stream Red Hat TZ=America/Indiana/Indianapolis _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx
