Firefox with your CLI it would instead execute the command to install it as snap. Shit like that is just outright disrespectful to the user.

I get it and, at the same time, I get it.

Ubuntu needs to be able to deliver with some level of guarantee for its corporate clients, which means testing. A browser like Firefox has a lot of dynamically linked libraries. How do ensure that it works with all reasonable combinations? 10 libraries with 2 supported versions each is 1024 combinations. A browser will have more libraries and more compatible versions of each, which leads to a massive number of combinations. Nothing like having a support customer with issues because a very specific patch version doesn’t work with another very specific patch version.

Compare that to snap. 1 artifact that contains all dynamically linked libraries. 1 artifact to test and support.

So, now Canonical has a tested and supported snap for a security sensitive application, whose method of delivery also isolates it from the host it runs on. Should they point users to that? Or some upstream binary that may have the above compatibility issues and lacks isolation, and wasn’t tested by them.

Short of it is that DLLs made a lot more sense when storage was expensive and programs were smaller. Now, they are problematic. Containers are a way to address that without having to update a ton of software, and they also improve security. If they hadn’t done it, the community would have torn them a new one for keeping the good stuff for their corporate clients.

That said, there have been a lot of missteps. The inability to have a self-hosted store of snaps (this may have changed since I last checked) and improper packaging of apps like Steam are good examples of this. On the other hand, PCSX2’s 32-bit version ran just fine long after Ubuntu went 64-bit-only.

There is no greater joy than when two of your enemies go to war against each other.

You make some great points. If I may, I’d like to expand on them with an alternative perspective.

When investing:

• A reasonable target for a business is around 10% per year (i.e., each $1M in assets should generate $100k in profit).
• An aggressive target for a business is around 20% per year.
• A conservative target for personal investments is around 3% per year.
• A more aggressive target is around 6-8% per year.

(These figures are approximate but are close enough for the purposes of my point.)

If a farmer has $10M in land and equipment, then we’d expect to see at least $1M profit per year. This is on top of the money that would be earned as a skilled employee who works significant overtime.

In 2022 Canada, potato farming (the second most profitable kind of farming) saw an average revenue of $600k on $480k of expenses. All the investment in land, its preparation, and the business enables an average Canadian potato farm to make the salary of a Canadian senior software engineer. And the senior software engineer doesn’t have nearly the buy-in costs.

Then there is the matter of risk. In Alberta, we’ve recently had some droughts that resulting in harvests so poor that harvesting what did grow was done at a loss. It’s like earning a paycheque so small that it’s not worth driving to pick it up. (The farmers did harvest the crops because it was necessary to collect insurance.)

So, yeah. High cost of buy-in, a lot of work, and a lot of risk for for the opportunity to make less than an American SWE.