Hosting your own LLM is anything but free. Aside from the constant operational expense with people monitoring and fixing issues, you need to provision enough resources and run your own inference server, which is both nontrivial and likely to perform far worse than OpenAI. There's legitimate reasons to host an LLM yourself, but it's not a "make this cheaper" button.
There may be a tipping point where you're burning XXM/year in API costs and the maintenance cost of rolling your own can be justified.
In the short term I agree, and one thing to consider is how rapidly the space is evolving and whether your team can even keep up with the latest advancements.
However, there will come a time when the bill comes due after launch and it will be very tempting to hire people to reduce the CapEx on the API.
Blue text color and underline are two things that universally help people recognize links – so no matter how custom your CSS is, using any of these traits for other purposes is a bad idea.
Of course, nowadays links can be any color and without underline, and we're getting used to it too. This only means that using any color to emphasise words in a text is a bad idea, as more and more people would see that as a link and get confused.
Web applications likely prefer buttons as links, which probably makes sense. They aren't "web pages" they are "applications". Text web pages stick to traditional HTML metaphors.
The typical case is that I'm testing with client code on my desktop that expects to talk to some particular server, such as db.work.com, on some particular port, say 3306. I don't want to run the client in a VM or container.
I want to run a test version of that server on a VM or in a container, and have that client connect to it, but I do not want to modify the client.
So I want to make an /etc/hosts entry for db.work.com giving the IP address of the VM or container that I'm running the test server in.
That works great with VMWare Fusion. The VM gets an IP address on my Mac, and I use that in the /etc/hosts entry.
That would also work great if I were running Linux instead of Mac OS, because Docker containers on Linux get IP addresses that are visible. I'd just have to put the container IP on the db.work.com /etc/hosts entry.
On Mac Docker runs a Linux VM and then the containers run in that. They don't have IP addresses that are visible to the Mac. For a lot of things that is fine. With the -p argument you can arrange to have a localhost port mapped to some port on the container.
But in my case that doesn't quite cut it. The client wants to connect to port 3306. I can't just map localhost:3306 to the container and put db.work.com in /etc/hosts pointing to 127.0.0.1 because I've already got something on 127.0.0.1 that is using 3306.
Hence Wireguard so that I can have an IP address for the container that is visible on the Mac.
They did infact
removed polkit : https://github.com/mozilla-mobile/mozilla-vpn-client/pull/70...
refactor auth using D-Bus: https://github.com/mozilla-mobile/mozilla-vpn-client/pull/71...
These are why author's PR was dropped.