jaeger changed the topic of #crux to: CRUX 3.7 | Homepage: https://crux.nu/ | Ports: https://crux.nu/portdb/ https://crux.ninja/portdb/ | Logs: https://libera.irclog.whitequark.org/crux/
jason123onirc has quit [Ping timeout: 240 seconds]
jason123santaoni has joined #crux
jason123santaoni is now known as jason123onirc
joe9 has quit [Quit: leaving]
<brian|lfs> hello is anyone around by chance?
<brian|lfs> I'm hitting my head on a brick wall I'm trying to make my own llvm package that also compiles rt-compilet and clang and I'm stumped on the actual extracting of all the files
<brian|lfs> its just not finding the source files at all that have already download
<brian|lfs> so cleary I'm missing something
<brian|lfs> here is my Pkgfile https://pastebin.com/u047NYeJ
<brian|lfs> also probably more elegant way of doing the same thing I would suspect.
<jaeger> Haven't tried to do so myself but do you NEED to override unpack_source() for this? Is there overlap between the source tarball directories or similar? If not, might be easier to avoid that
<brian|lfs> so how do I in the unpack section to it to extract a file from PKGMK_SOURCE_DIR
<brian|lfs> to $SRC
<brian|lfs> I see thats my issue now
<jaeger> Ah, you mean your tar commands are not finding the tarballs in unpack_source() ?
<brian|lfs> exactly
<jaeger> Try using PKGMK_SOURCE_DIR, see contrib/go/Pkgfile for an example
<brian|lfs> thanks looking
<jaeger> The destination will be in $SRC but the tarballs downloaded should be in $PKGMK_SOURCE_DIR
<jaeger> Sure
<jaeger> Side note, the && on lines 37 and 38 doesn't do anything useful
<jaeger> Same on the later lines, if they're not also ending with \
<jaeger> 51, 54, 56, 57
<jaeger> etc.
<brian|lfs> understood
<brian|lfs> thanks I probably went through all of this before then a new version of LLVM came out and I saw that my patch wasn't working in the past
<jaeger> I haven't tested it but maybe something like this: http://ix.io/4Imj
<brian|lfs> well it 99% worked on LLVM-16.0.5
<brian|lfs> its just I forgot to tell it where the patch was located. and feared that I had compeltely screwed up and started over with a fresh start
<jaeger> ah
<brian|lfs> compiling
<brian|lfs> it could of been missing clang and or rt before hard to say
<brian|lfs> but it didn't cause anything to break on me
<jaeger> Good luck, I'm out for the night
<brian|lfs> ah almost 1am already
<brian|lfs> jager did we drop wget-paste from crux I can't find it anywhere a Pkgfile for it?
<brian|lfs> If anyone wants or needs wgetpaste https://bpa.st/KTUQ
<brian|lfs> not sure why CRUX had no port of it
sajcho has joined #crux
sajcho has quit [Client Quit]
sajcho has joined #crux
<sajcho> brian|lfs: It might help https://github.com/sajcho/opt-x86_64/blob/main/llvm-toolchain/Pkgfile - it's just a suggestion
<brian|lfs> thanks I'll look at it at somepoint mine worekd after figuring out extracting of all the tar fiels
<brian|lfs> wow I can't type
sajcho has quit [Quit: Client closed]
<SiFuh> brian|lfs: Looks like that was typed by me actually.
<brian|lfs> what the wgetpaste or the llvm?
<SiFuh> worekd fiels <-- The line with these
<SiFuh> At first I thought it was me but then I realised it is probably the wireless keyboard/USB dongle or USB port. I have to move the keyboard closer to the dongle or letters get mixed up or missing.
<SiFuh> Gets worse when the batteries are nearly flat
<brian|lfs> ok looked at your package file so does it have rt-compiler enabled also?
<brian|lfs> or just clang
<brian|lfs> oh it does says it does in the description
<brian|lfs> do you still run CRUX or is it your own custom thing like me SiFuh?
<brian|lfs> I haven't figured out what to call my work yet I'm baseing it on LFS multilib and systemd
<SiFuh> brian|lfs: Only on the repo sever. Everything else is OpenBSD
<brian|lfs> ah ok OpenBSD Linux kernel or BSD kernel?
<SiFuh> OpenBSD kernel
<brian|lfs> anything BSD I always had lack of hardware support
<brian|lfs> I'm sure a 13900k would puke
<cruxbot> [contrib.git/3.7]: python3-bleach: 6.0.0 -> 6.1.0
<cruxbot> [xorg.git/3.7]: xkeyboard-config: 2.39 -> 2.40
<cruxbot> [core.git/3.7]: cmake: 3.27.6 -> 3.27.7
lavaball has joined #crux
joacim has quit [Ping timeout: 260 seconds]
joacim has joined #crux
<cruxbot> [contrib.git/3.7]: aml: initial commit, version 0.3.0
<cruxbot> [contrib.git/3.7]: neatvnc: initial commit, version 0.7.0
<cruxbot> [contrib.git/3.7]: wayvnc: initial commit, version 0.7.1
<cruxbot> [contrib.git/3.7]: wasi-libc++: 17.0.1 -> 17.0.2
<cruxbot> [contrib.git/3.7]: wasi-compiler-rt: 17.0.1 -> 17.0.2
<cruxbot> [contrib.git/3.7]: wasi-libc: 20230918 -> 20231003
farkuhar has joined #crux
r0ni has quit [Remote host closed the connection]
r0ni has joined #crux
tarxvfz has joined #crux
<cruxbot> [core.git/3.7]: libtirpc: 1.3.3 -> 1.3.4
<ukky> Are there any advantages/disadvantages of having multiple single-applet busybox binaries, versus single multi-applet busybox binary combined with per-applet symlinks?
braewoods_ has quit [Read error: Connection reset by peer]
<SiFuh> ukky: "BusyBox benefits from the single binary approach, as it reduces the overhead introduced by the executable file format (typically ELF), and it allows code to be shared between multiple applications without requiring a library."
lavaball has quit [Quit: lavaball]
<cruxbot> [compat-32.git/3.7]: libtirpc-32: 1.3.3 -> 1.3.4
<ukky> SiFuh: What happens during hotplugging when only mdev is in use? If there are thousands of events, does system keep image in memory in-between hot-plug events? Or busybox image gets paged-out in-between events?
<SiFuh> Not a question I can answer. Someone else may know here
<ukky> Having single multi-applet busybox binary is great imho, but I just worry about deteriorated mdev functionality. Unless it's not paged-out, then it is fine for mdev too.
<SiFuh> I remember reading about the topic of your first question a while back when I was looking at making a hybrid CRUX install CD where all I needed was to copy the ports over rather than doing the full ISO script build. But that is as far as I got.
<SiFuh> I didn't like that the rootfs for CRUX was quite large these days and takes awhile to unpack and uses more than 1GB of memory. So my plan was to make a USB stick with two partitions. One for busy box and one for the CRUX ports and the setup scripts
<ukky> That is doable.
<SiFuh> I then was poking around 'SeaTools Bootable' for seagate hard drives and then I can't remember why I stopped.
<ukky> At work I remember combining Coreboot and SeaBIOS to make system boot from SATA/ISO/USB
<SiFuh> SeaTools had a USB version and I think it was Puppy Linux or Damned Small Linux. It was super fast
<ukky> Are there many distros that beat CRUX on minimum memory requirement?
<SiFuh> As in CRUX installed or CRUX ISO
<SiFuh> CRUX installed is really good with memory and pretty close to the other distros with minimum requirements.
<SiFuh> The ISO on the other-hand is nowhere near the bottom of the list near minimum requirements.
tarxvfz has quit [Quit: tarxvfz]
<ukky> SiFuh: my question was about bootable ISO. Do we need/want minimalist bootable ISO?
<SiFuh> I think we do, but it isn't up to me though and who knows what the future will hold
<ukky> If we need tiny bootable ISO, we need to come up with the list of minimum packages for the rootFS and also minimum list of features the ISO should have.
<brian|lfs> SiFuh, I meant by puke on open BSD unsupported hardware.
<SiFuh> brian|lfs: AMD and Radeon
<SiFuh> Works very well
<brian|lfs> ok but with modern hardware
<SiFuh> Yeah, only thing that doesn't work is the Wi-Fi 6 from Mediatek I think it is
<SiFuh> "MediaTek MT7921K" MediaTek Inc. Wireless_Device
<cruxbot> [contrib.git/3.7]: python3-blinker: 1.6.2 -> 1.6.3
<cruxbot> [contrib.git/3.7]: python3-aiohttp: 3.8.5 -> 3.8.6
<cruxbot> [opt.git/3.7]: shared-mime-info: 2.2 -> 2.3
lavaball has joined #crux
armin has left #crux [#crux]
<brian|lfs> I've run into an odd bug either with pkgmk or curl?
<brian|lfs> if it tries to pull a package from a url and the file name has both capital and lowercase it errors out if you use the $name variable
<brian|lfs> try substituting with $name for the file name and it doesn't download.
<brian|lfs> example if someone wants to try and duplicate https://bpa.st/4LJA
<farkuhar> brian|lfs: replacing kImageAnnotator with $name worked for me, http://ix.io/4Iqk . Maybe there was a temporary problem with the Github url redirector.
<brian|lfs> interesting
<brian|lfs> what package does pkgmk come with I could see what version I have installed?
<brian|lfs> I'm on 5.40.10
<brian|lfs> for pkgutils
lavaball has quit [Remote host closed the connection]
<farkuhar> brian|lfs: yes, 5.40.10 is the current version of pkgutils. There's no reason for pkgmk to behave any differently using $name instead of kImageAnnotator. If github redirection is not broken, then maybe your first guess (curl bug) is on the right track.
<brian|lfs> not a huge deal just happened to be something I bumped into
<farkuhar> if you don't use the same capitalization when naming the port directory, though, prtverify will give you an "ERROR ... variable name do not match the port name".
tilman has quit [Ping timeout: 272 seconds]
tilman has joined #crux