<SiFuh_> farkuhar: https://git.crux.nu/tools/rc/pulls/2 <-- Very funny.
<SiFuh_> https://git.crux.nu/farkuhar/rc/src/branch/master/rc <-- This is bullshit too. Why the fsck does anyone want colour in the rc itself?
<SiFuh_> printf "%b" "${COLOR_ERR}"
<SiFuh_> printf "%b\n" "${RESET}"
<SiFuh_> You can do that better as well. Get rid of these things
<SiFuh_> And what is the point of setting colors to false when rc.colors can't be found? You are already defining colours in the rc
<SiFuh_> So get the colours and the variable for colours out of rc completely Put it in a sourced file
<SiFuh_> Apart from that, everything else looks good and nice and neat
zorz_ has joined #crux-musl
<SiFuh_> nd as a side note farkuhar. This rc script with your colour addition will not work on old terminals
<SiFuh_> farkuhar: Line 103. You missed something important. vgscan for LVM. This was key to solving much of the issues in CRUX.
<SiFuh_> farkuhar: rc.shutdown should be before services shutdown.
<SiFuh_> Boot goes services -> rc.local shutdown should be shutdown.local -> services.
SiFuh_ has quit [Remote host closed the connection]
SiFuh_ has joined #crux-musl
SiFuh_ has quit [Quit: leaving]
SiFuh_ has joined #crux-musl
SiFuh_ has quit [Client Quit]
SiFuh_ has joined #crux-musl
<farkuhar> SiFuh_: added back the vgscan command with the latest commit. Also updated the man-page to warn about unprintable control codes on older terminals.
<farkuhar> The printf commands on either side of the full-screen fsck error message have been protected behind a [ "$USE_COLOR" = true ] test, so they won't be activated if the user disables colour.
<SiFuh_> farkuhar: defeats the point.
<SiFuh_> Colours and variables should be completely removed from rc and a separate rc.colors script compliments the rc script.
<SiFuh_> farkuhar: Are you against the idea of using awk xargs and sort in the rc?
<SiFuh_> Because that umount command could be sorted to umount in order rather running umount multiple times to umount everything.
<SiFuh_> Maybe you know of a bash way to do it
<farkuhar> SiFuh_: I'd prefer not to rely on awk xargs and sort. What if awk and sort happen to break, after a shared library update? If bash can run, though, then launching the two back-to-back umount commands shouldn't be an issue.
<SiFuh_> And a manual edit isn't going to fix the issue. Don't put colours in rc. Simple. Then you can use the ideas from my rc.functions to support both modern and older terminals.
<SiFuh_> And yes, a separate port is the best option.
<farkuhar> You were probably on the right track to suspect that the removal of the second umount command was what led to journal corruption of my /home partition. I wish there were a way to avoid the two back-to-back `umount -a -r` commands.
<SiFuh_> Well it shouldn't really have effected home. If umount umounts only that which can be and errors out that which can't, then reboots. The only way I see it happening to /home is if /home has something mounted under it or /home is actually mounted two levels under /
<farkuhar> Nope, /home was mounted directly under / in my case.
<SiFuh_> But yes, I looked into it last night and tested it. First umount would error for / /usr/ but it would have umounted /usr/local. The second command would error because it would not be able to umount / but umount /usr. However / would already be readonly and so would have /usr from the first umount command.
<farkuhar> Anyway, the fsck error message draws enough attention to itself being full-screen. There's no need to colorize it any further, so I can just remove the printf lines surrounding the heredoc.
<SiFuh_> farkuhar: Also, have you shifted your ridiculously long rc.shutdown.local upwards?
<farkuhar> Oh, the local shutdown script was moved to the proper location too.
<SiFuh_> farkuhar: Yep as I said, what is the point of colour for it? You need binoculars to read your screen to notice it?
<farkuhar> On a 4k screen with FONT=default? Maybe.
<SiFuh_> No, you see it
<SiFuh_> The error box is about the size of a business card on 4K
<SiFuh_> By the way, why did you remove the printf entering multi-user mode line?
<SiFuh_> /bin/stty onlcr
<SiFuh_> I saw that and forgot to put it in.
<zorz_> yeaaahh farkuhar is to snob to chat in #crux-social now
<SiFuh_> Yeah, he's all royalty now.
<zorz_> only when he needs something :P
<SiFuh_> farkuhar: Now you are calling a static variable multiple times.
<SiFuh_> constant*
<zorz_> actually this is good... i am not going back to runit.
<SiFuh_> printf "\033[1;37m%s:\033[0m %s\n" "$1" "$2" <-- I'd get rid of this too now. As I said. Use your sourced rc.colors
<SiFuh_> The terminal I use has no styles. Only can use colour. so 1;37 will not work. only 37 will
<SiFuh_> And in beermans script he didn't even bother to put colour for that. he just made it bold.
<farkuhar> That's for printconf(), right? Okay, the control code for bold must be eradicated.
<SiFuh_> What I am saying is put all colour into your rc.colors. Leave no entry for colours at all in the rc. No variables. The USE_COLOR should be accessed by rc.colors and rc sources it.
<SiFuh_> That means you can easily make an rc-color-mod port and the original rc stays the same regardless. Only entry you should have in to source rc.colors
<SiFuh_> Only entry you should have in rc is to source rc.colors
<SiFuh_> VERBOSE can stay in the rc
<SiFuh_> It will print plain text if the user doesn't want color but wants verbose mode
<farkuhar> VERBOSE is only tested once in each script, right?
<SiFuh_> rc.colors can overide the default printconf, printinfo and prinrerror that is in the rc if you source it from rc.colors.
<SiFuh_> VERBOSE is test once. But I put it in rc.functions just in case rc.colors didn't get it from rc.
<SiFuh_> rc.colors/rc.functions heh
<SiFuh_> I prefer rc.colors.
<farkuhar> Regardless of the name, it's not meant to be executed standalone. So if it's only being sourced by rc (not executed), then the shebang doesn't really matter.
<SiFuh_> farkuhar: Yes. But perfection is the key.
<SiFuh_> I wonder if bash reads that shebang and understands it as a functioning script
<SiFuh_> cat << EOF for the massive failure scary box thing. Isn't that how it was done back in the early days?
<SiFuh_> You could probably assign a variable to it or make it a function and use rc.colors to catch it.
<SiFuh_> I still don't know why the devs let olutmies, a port maintainer get away with butchering the rc.
<zorz_> who is olutmies?
<SiFuh_> zorz_: Ask ppetrov^
<zorz_> miller lite?
<SiFuh_> beer man
<zorz_> yeaaa
<zorz_> this boy should stick with contrib
<SiFuh_> He should go back to ubuntu
<zorz_> like this one
<zorz_> source=(https://github.com/Duncaen/OpenDoas/releases/download/v$version/$name-$version.tar.xz doas.pam better-wayland-support.patch)
<zorz_> wayland support
<zorz_> i dont do wayland?
<SiFuh_> Pam sucks. I get slow login with it
<zorz_> i am fast
<SiFuh_> Yeah I know, the ladies say that about you all the time.
<zorz_> my firefox, now is really fast, pops up almost like a terminal
<SiFuh_> zorz_: I was looking at that. But got carried away with writing farkuhar's rc script
<zorz_> without libdav1d and some other bullshit
<zorz_> xorg-libxt and this has to be removed
<zorz_> wait get the new one
<zorz_> ac_add_options --disable-webspeech
<zorz_> it always pops up i dont have support... fuck you and your webspeech
<zorz_> :P
<zorz_> let me remove and the optional
<zorz_> SiFuh_: basically is farkuhar' build.... i just changed it :P
<SiFuh_> zorz_: That's not how it works. You are suppose to only credit him for one tiny thing, and you mess it up. And everything else he did, you don't mention his name at all so it appears you did the entire thing
<zorz_> ofcourse i will credit him
<zorz_> i will change the maintener
<SiFuh_> Hehe
<zorz_> you know i dont use sccache, lld, c-ares, gnutls and and and
zorz_ has quit [Quit: leaving]
<farkuhar> Hehe, olutmies. For a minute there I thought it was another misspelling of uwumeowmeownyaa (payback for deliberately misspelling CRUX). Now if I actually spoke Bulgarian, the confusion might not have arisen.
<farkuhar> We don't really need another misspelling of uwumeowmeownyaa. Osas is enough. Keep it short and simple.
<farkuhar> It was never really "my" rc script. I only served as the delivery boy, using my existing Gitea credentials so that the other collaborators wouldn't have to create a new account for something the core team screwed up. rc is in the hands of the wider CRUX community now, but especially the core team who should be picking through our pull request with a fine-tooth comb.
<farkuhar> zorz_ shouldn't bother to credit me for anything I did to the firefox port. I inherited a solid base from Romster, and only tinkered around the margins as I discovered subtle edge cases that the existing build() didn't handle.
<SiFuh_> Heh, I was teasing you. The problem is beerman thinks I did nothing. But good thing you didn't yet give me credit because your additions were a bit.... well.. I don't want my name on colour and variables for color in rc. But anyway
<SiFuh_> I will probably tease you a lot more though in the future about it.
<SiFuh_> farkuhar: When you going to upload the rc.colors? Any estimated time frame?
<farkuhar> SiFuh_: upload where? As a standalone contrib port?
<SiFuh_> farkuhar: Hmmmm.. Good point.
<SiFuh_> Yeah, make it into a port. opt or contrib? opt right?
<farkuhar> Okay, opt will be a good home for it.
<SiFuh_> Cool! This can please both beerman and everyone else
<SiFuh_> Hahaha
<SiFuh_> farkuhar: By the way. When I was writing the rc script I noticed single and shutdown run the same service shutdown script. That could be made into a function in the future
<farkuhar> Yes, it could.
<farkuhar> I'm about to push an update to contrib/python3-lxml (do not force a bytecode level), but my only machines available for testing it are running CRUX-MUSL. I hope the hand-edited footprint will not cause any mismatches on CRUX-glibc.
<SiFuh_> Heh
SiFuh has quit [Remote host closed the connection]
SiFuh has joined #crux-musl
<SiFuh_> iwd hehehe It was dumb ;-)
zorz_ has joined #crux-musl
zorz_ has quit [Quit: leaving]