<farkuhar>
SiFuh_: zorz is willing to put his laptop to work building an ISO. But if he tries to follow this procedure https://gitlab.com/SiFuh/Documentation/-/blob/master/Musl/CRUX-MUSL.txt?ref_type=heads he will have to improvise, considering all the changes we've discussed since then (runit alongside sysvinit, for example). Perhaps a revised procedure is needed, one that zorz can perform without much help.
zorz has quit [Quit: leaving]
zorz has joined #crux-musl
SiFuh is now known as SiFuh__
SiFuh_ is now known as SiFuh
SiFuh__ is now known as SiFuh_
zorz has quit [Quit: leaving]
zorz has joined #crux-musl
zorz has quit [Quit: leaving]
zorz has joined #crux-musl
zorz has quit [Quit: leaving]
zorz has joined #crux-musl
SiFuh_ has quit [Remote host closed the connection]
SiFuh_ has joined #crux-musl
zorz has quit [Quit: leaving]
zorz has joined #crux-musl
zorz has quit [Quit: leaving]
zorz has joined #crux-musl
zorz has quit [Ping timeout: 260 seconds]
zorz has joined #crux-musl
<SiFuh>
dlcusa: If you want. Are you going to do a MUSL system?
<SiFuh>
Then if so, you can use ukky's ISO as the foundation.
<dlcusa>
I did think MUSL was the most important goal of this effort.
<SiFuh>
dlcusa: I was building a glibc version
<SiFuh>
serpente wants a multilib version
<dlcusa>
Does that make sense for this project; i.e., who will be booting it for what purposes?
<SiFuh>
And opendoas and sudo goes to opt not core.
<SiFuh>
I see nothing wrong with using musl, except those who use games on Linux may have some issues.
<dlcusa>
Well, let's do that later, then. I don't expect all CRUX users to play with this for a while.
<SiFuh>
Well reverting to glibc is easier than going to musl. Because we already have a glibc foundation. That is why emmett1 and I did the 3.7 musl so we'd have a foundation. And ukky has already made the 3.8 foundation.
<dlcusa>
What is your recommendation, then?
<SiFuh>
Go MUSL. I can adapt it later to glibc
<dlcusa>
Noted.
<SiFuh>
dlcusa: I am more interested in my ANSI C rc at the moment
<dlcusa>
I agree with your prioritization.
<SiFuh>
dlcusa: I did it for fun, but damn it boots insanely fast.
<dlcusa>
That IS one of the major objectives of CRUX (even faster than systemd!).
<SiFuh>
No need for colours. You blink and the system has booted. :-P
<dlcusa>
I think colorization is a really low priority, myself, but my bread IS pretty grey.
<SiFuh>
I still have rc.local and rc.conf and stuff. But the main rc is all ANSI C
<SiFuh>
The source files are easily modifable. You just recompile it.
<dlcusa>
I do colorize my shell prompt.
<SiFuh>
Same. But we look at that longer than a system booting.
<farkuhar>
dlcusa: grey bread or grey beard?
<SiFuh>
farkuhar: A hazy shade of grey
<dlcusa>
Precisely. Grey and gray are completely interchangable.
<farkuhar>
But grey on bread might be a sign of mold. Whereas a grey beard merely adds an air of respectability.
<SiFuh>
Grey is the British spelling. Gray is the US spelling.
<dlcusa>
s/bread/beard/ -- my typing's not what it used to be.
<SiFuh>
farkuhar: He want's multilib for playing games
<farkuhar>
Blindly trying to exchange x86_64 for <arch> throughout the footprint might have unwanted side-effects. Similarly, exchanging "gnu" or "musl" for <libc> throughout the footprint will sometimes do more than what you want.
<SiFuh>
dlcusa: Oh and I built for libressl and not openssl
<dlcusa>
Is there a consensus on libressl going forward?
<SiFuh>
farkuhar: Also for the default editor. I see CRUX has VIM
<SiFuh>
There should be a slimmed down vi and not vim
<SiFuh>
dlcusa: No idea. Have to ask the users to vote on it. But you will run into future issues when compiling some ports. But generally it works for most ports.
<dlcusa>
Concur--vim default is not aligned with CRUX philosophy, IMHO.
<SiFuh>
dlcusa: I have mine as /usr/include/libressl but if it is to be the default, it should probably be /usr/include
<farkuhar>
I think pitillo had a better approach to pkgutils PR#2, actually. Generate multiple footprints, with architecture (or libc provider) as a suffix. Then the pkgmk check_footprint() function could select the right footprint for comparison, based on the machine where it's running.
<SiFuh>
dlcusa: I have emacs_no-x11 :-P
<dlcusa>
We hold this truth to be self-evident that vi is more CRUX than vim.
<SiFuh>
Yes, and I don't mind that myself.
<farkuhar>
SiFuh: I agree, vim is too bloated for inclusion in core.
<SiFuh>
vim and emacs should be in opt. In fact, one or both use jannson
<SiFuh>
nftables too. And nftables should be in opt
<dlcusa>
Hey, we're not preferring ed over vi.
<SiFuh>
Mostly because some uses prefer iptables and some prefer nftables.
<farkuhar>
If we did generalize the pkgmk check_footprint() function so that it took into account the CPU architecture and libc provider, we shouldn't require the port maintainer to generate the corresponding footprint for all possible combinations. The downstream user can benefit from check_footprint() if they have the same CPU and libc provider as the maintainer, otherwise they just generate their own footprint after a successful build
<SiFuh>
farkuhar: I still think build utils should have a seperate repo. That way they can easily be disguished from real core ports.
<SiFuh>
If I build a server, I would like to know what I don't need on my system anymore.
<SiFuh>
That includes pkg and prt utils.
<SiFuh>
pkgutils maybe in core though
<dlcusa>
It looks like I need free gitea to get access to everything I should be able to investigate, am I right?
<SiFuh>
I don't have it.
<farkuhar>
dlcusa: What do you anticipate having to investigate?
<farkuhar>
The repos on git.crux.nu that you would need are all publicly accessible, no account needed to view them. Opening a PR or an issue does require an account, though.
<dlcusa>
Then that's one less thing to spend time learning. :-)
<farkuhar>
What I need to learn about Gitea is whether it's possible to rebase an existing PR, once my commits have diverged from the commits on the corresponding branch of the canonical repo. I think I might need to `git reset --hard` and apply our revisions onto a new branch, forked from jue's last commit.