Profile: 28b1b69e...

nostr:npub19zcmd845ct95g9q487mr02jzwuhzg4f0knz33prsaq7lw2vcvqxqt5padn Yeah and while distros can get embargos for OpenSSL so they can at least prepare themselves, you're likely not going to.

And OpenSSL is used for HTTPS requests, so pretty much directly network-facing and it can have rather nasty security vulnerabilities.

nostr:npub1ysufjjd485tftr4wy2a83fqyqvtfq0yn820gl8vl6hcsdz8uv2hskx2jyl Would an option to do "static except these specific things" make sense? :blobfoxthink: Or are we just back at square one then? Because if OpenSSL can be such a problem, I assume distros make sure they have the newest version any how, so it wouldn't create the compatibility issues we see now. Or am I seeing things too simplistic?

nostr:npub19zcmd845ct95g9q487mr02jzwuhzg4f0knz33prsaq7lw2vcvqxqt5padn Even if this would exists, I wouldn't do it as one of the dependency is OpenSSL (pulled by Erlang).

nostr:npub1ysufjjd485tftr4wy2a83fqyqvtfq0yn820gl8vl6hcsdz8uv2hskx2jyl what's the problem with that? Is it that OpenSSL needs to be easily/more quickly updateable?

is there any one smart enough to file an issue somewhere to get the Elixir release builds be static builds? So we don't need all these flavours any more and people can just drop the binary on w/e distro they have as long as the cpu architecture is good. (As I understand, that should be possible with static builds, right?)

I guess having dynamic builds makes sense for corporations who control their infra, but for floss projects, that just seems like a bother in a lot of cases...

I'm not smart enough. I don't know where to ask, how to ask, and if they want more information, I'll probably not even understand the question :/

a hack i've been thinking of i also want to try someday.

When your instance goes down, you get annoying 500 errors, right. But what if you can set a retry in the http proxy and tell it to retry for several seconds. You could basically restart while it looks like you still have uptime, hehe (just bigger latency).

And a quick glance tells me something like that may be possible in Caddy by using the load balancers lb_try_duration and lb_try_interval. And yes, this is load balancing settings, but who's to say you can't have a load balancing setup with only one node :blobfoxcomfysmirk: