There are big wishes for Signal to adopt the perfectly working Flatpak.
This will make Signal show up in the verified subsection of Flathub, it will improve trust, allow a central place for bug reports and support and ease maintenance.
Flatpak works on pretty much all Distros, including the ones covered by their current “Linux = Ubuntu” .deb repo.
To make a good decision, we need to have some statistics about who uses which package.
I prefer the deb that works. I get a signal.update almost every other day. I don’t remember to update my flatpaks anywhere near that often. I also appreciate that it doesn’t force me to include dependencies that are already met.
You can update flatpaks automatically using systemd. Automatic updates are a thing and should be everywhere.
https://discuss.kde.org/t/improving-metered-network-detection-and-usage/9287
Absolutely not…most especially prior to production deployment. How else would someone see the change logs before hand or see/test if it would hurt their environment?
I have no idea what a production environment is for you. If it is some kind of sealed off stuff yeah maybe, but otherwise I hope you use a Distro that handles updates the way you need it.
Not updating because things will break is a sign of a bad distro.
Production environment is typically in the corporate world, not usually a homelab. Service providers often have a SLA uptime guarantee of 99%. They don’t often push patches as soon as available due to the varied nature of corporate environment. They don’t have one or two PCs to worry about: they can have tens of thousands. Downtime equates to money lost. So patches get tested before being deployed. Depending on the patch, that can be 48 hours to a week or two. Major OS upgrades can be months-long test, but the company usually does that and follows it while it’s still in beta.
Updates are pointed to a server the company controls, not the Internet. Updates get tested on test servers and test machines that replicate those in production. It typically gets monitored for 48 hours to measure glitches and performance. Once satisfied, the company controlled update server pushes into production machines.
Why test patches before deploying to productions?
Ok this is a specific case, interesting info, thanks.
Obviously this has nothing to do with single user computers, going to software stores, pressing “update” buttons etc.
So it is unrelated to my point.