serpente has quit [Remote host closed the connection]
<zorz>
q
<SiFuh>
dlcusa: Yes exactly like that ;-)
<SiFuh>
remiliascarlet: zorz: farkuhar: ukky: I was thinking to do something like this. Save it as setup-kernel on the ISO. https://dpaste.com/DKD742XFB.txt
ppetrov^ has joined #crux-social
<SiFuh>
ppetrov^: remiliascarlet: Asked ChatGPT a question. "If I had a 70 KG prism being lifted by two straps. Each end was connected to a D-Shackle then what would be the weight on each shackle if distributed evenly?"
<SiFuh>
So it does this. "70kg x 9.8m/s^2 = 686 N. 686N / 4 = 175.5 N. 9.8m/s^2 / 171.5N = 17.5 kg" and I asked it "Would it not be easier to just divide 70 kg by 4 to get 17.5 kg?"
<SiFuh>
"Haha, you're absolutely right! 🤣 It’s funny how we got all technical with newtons when, for your original question about how much force each D-shackle holds, we could have just simplified it by dividing the weight in kilograms by the number of shackles."
<SiFuh>
We? It did it. Not me.
zorz has joined #crux-social
<farkuhar>
Can you believe I've never yet interacted with ChatGPT directly? My only experience with chatbot AI has been vicarious, by way of reports like SiFuh's.
<farkuhar>
For those of us who don't have base64 and WebP decoders in our heads, here is the image that dlcusa tried to share: https://d.uguu.se/WaOiENLt.webp
zorz has quit [Quit: leaving]
<farkuhar>
SiFuh: https://dpaste.com/DKD742XFB.txt seems to run the make commands from *outside* the chroot (i.e., in the live ISO environment). Does your live ISO provide a working gcc? If you don't need to chroot to the newly-installed system to compile the kernel, then the handbook can be simplified.
<SiFuh>
farkuhar: You don't need to be in a chroot when building the kernel from the ISO installation
<SiFuh>
farkuhar: But it wouldn't be hard to chroot it into it and remove /mnt from everything.
<SiFuh>
dlcusa: Did you see the South West Airlines and the Private Jet incident?
<farkuhar>
SiFuh: ukky: I just performed an upgrade to CRUX-MUSL 3.8 using ukky's ISO. Reboot was successful, and everything seems to be working so far.
<ukky>
farkuhar: nice. Did you have 'linux-headers' in 3.7? If not, there might be a collision, depending on update sequence.
<farkuhar>
setup-helper on the ISO should be revised for MUSL, removing libxcrypt and libxcrypt-32 from the inject_packages line.
<farkuhar>
ukky: The setup script runs pkgadd with -f (force), so it wouldn't complain about overwriting files owned by another package. But yes, I did have linux-headers installed on 3.7
<farkuhar>
The prt-utils repo has a commit from last August (revdep: do not try to read ld.so.conf on non-glibc systems) but I haven't bumped the version or release in the port itself (not even on the 3.8 branch). Maybe I should bump the version so that the final 3.8 ISO has a working revdep.
<farkuhar>
As for the possible collision with linux-headers, you could add some lines to setup-helper, removing any existing linux-headers package before the upgrade.
<SiFuh>
farkuhar: ukky: touch /etc/ld.so.conf
<farkuhar>
SiFuh: I'll try your setup-kernel.sh from the live ISO, but I distinctly recall that `which gcc` returned nothing when I was in the live ISO environment. Even if your script exported PATH=$PATH:/mnt/usr/bin before trying `make menuconfig` and `make all`, that might not be sufficient to get a working toolchain.
<ukky>
Setup ISO has no gcc
<ukky>
I had a development ISO with full toolchain+git on the ISO (for crux-3.7-glibc)
<SiFuh>
I'd just need to append a section where it goes into a chroot. But the problem is if they ran setup-chroot then it the script would need to run chroot only. If the you never ran setup-chroot then the script would need to setup-chroot.
<farkuhar>
Maybe SiFuh builds his ISO differently. He saves so much space by not including any packages from opt and xorg, he might as well provide a working gcc on his ISO.
<SiFuh>
farkuhar: So I am still building an ISO then?
<ukky>
Hahahaha
<SiFuh>
fuckinguhar
<farkuhar>
Sorry, I should have used the past tense. The ISO that SiFuh and emmett1 built for CRUX 3.7 was compact enough (being core-only) to allow the inclusion of a working gcc, if they wanted to allow users to build the kernel without chrooting first.
<SiFuh>
farkuhar: I fresh installed ukky's ISO and built the kernel and stuff. All went flawlessly.
<ukky>
Let's collect what we'd like to change in the ISO, I might re-run last ISO creation step that does not involve rebuilding packages: 1) setup-helper for musl; 2) SiFuh kernel build script???
<farkuhar>
Heh, I didn't bother to build a new kernel on the machine I just upgraded. I kept the old kernel (6.10.7) which I built last September, and didn't change anything in the syslinux config.
<SiFuh>
ukky: could always install ckut from the ISO if we included it. But you'd need a working internet, or i'd need to tweak it to detect what kernel is extracted on the system which would mean I can dump all the internet stuff
<SiFuh>
farkuhar: I did a format and I wanted to test zstd straight from the ukky
<SiFuh>
And I installed everything, which I don't want. But it is nice to have it as close as CRUX as possbile, but to be honest, a core and some basics in opt is usually enough.
<farkuhar>
SiFuh: Well it's good that we're all testing different ways to use ukky's ISO. You've tested the clean install path; I've just tested the upgrade path.
<SiFuh>
Yes
<ukky>
Do we want to patch every location where 'crux' name is mentioned and replace it with crux-musl?
<SiFuh>
So what is wrong with the gcc?
<SiFuh>
ukky: Yes, but it isn't much.
<SiFuh>
But I don't think it is important. It's just there for coolness and reminding them or others it isn't a glibc system :-P
<farkuhar>
Nothing wrong with the gcc; that's one of the few places where ukky *did* change the name to crux-musl.
<ukky>
SiFuh: gcc is not on live ISO, only available as tar.xz, you have to chroot to have gcc available.
<SiFuh>
If I remember correctly it is grub, syslinux, iso-release, crux, setup, setup-upgrade and filesystems
<SiFuh>
ukky: Ahh I see
<ukky>
Do we want to keep '/lib64' and '/usr/lib64' symlinks? But that change would require 'filesystem' rebuild
<SiFuh>
ukky: farkuhar: I'd like to modify ports with a script that convers everything /usr/* in core to / and everything in contrib to /usr/local
<farkuhar>
SiFuh: should be easy to accomplish with stenur's overlay_build feature
<SiFuh>
Oohhh
<SiFuh>
farkuhar: Could insert it into ports -u
<farkuhar>
And I misspoke about not hitting the issue with xorg-xf86-video-intel. It's my desktop that doesn't need the Intel GPU drivers; the CRUX-MUSL laptop does have that package installed. But thankfully no failures are surfacing yet.
<SiFuh>
There is a symlink fro /var and /usr/var
<farkuhar>
SiFuh: Don't get me started on ports -u. That script should be cast into the rubbish heap of history.
<ukky>
For core/opt/xorg collections, do you want to keep the names, or should I rename them to muslcore/muslopt/muslxorg when I create those repos?
<ukky>
For the issue with xorg-xf86-video-intel that SiFuh reported, it might be just 'ldd' script that cannot resolve undefined symbols.
<farkuhar>
ukky: How do we intend to make these repos available: git only, rsync+git, httpup+git, or rsync+git+httpup?
<SiFuh>
ukky: Maybe because it didn't log the error from the output of each lib. So it might be itself.
<ukky>
farkuhar: I want httpup+rsync
<farkuhar>
Ah, one of the combinations I didn't anticipate. Why exclude git?
<SiFuh>
core-musl muslcore ? Hmm might be a good idea
<SiFuh>
I use musl nopam core opt xorg contrib. But I have the opt that is modified for musl in musl as well and I find that annoying.
<ukky>
farkuhar: git will be more complex to setup on my VPS. httpup should work. rsync should work.
<farkuhar>
I just assumed we'd make them available through git, if we hosted them on the crux.nu Gitea instance.
<SiFuh>
core-musl musl-core coremusl muslcore ;-)
<SiFuh>
I think use the hyphen like compat-32
<SiFuh>
core-musl opt-musl xorg-musl
<ukky>
farkuhar: I can upload 'git diff' and you can add repos to crux.nu
<farkuhar>
ukky: I think when you fork an existing repo in Gitea, by default it keeps the same repo name. I should pay attention to see if Gitea gives you the option to change the repo name.
<farkuhar>
Hence you'll see multiple copies of official repos on the crux.nu Gitea instance, all under the namespaces of individual users: tb, jlngn, Leeroy, groovy2shoes, ...
<ukky>
farkuhar: The name should not matter, as all three collections are separate git repos. Thus all three diffs are relative to individual collection, disregarding repo name.
<farkuhar>
For the diff it doesn't matter, I agree. But SiFuh's question seems to be about branding, i.e., establishing a distinct identity for this fork.
<SiFuh>
ukky: farkuhar: Why core-musl is good. ls -l /usr/ports you will have core and core-musl in that order.
<SiFuh>
And when they cd core <tab> <tab> it will show both
<ukky>
ISO Makefile has to be patched heavily to accomodate new musl collection names
<farkuhar>
Heh, at least you didn't scroll the conversation 3+ screens by pasting a base64-encoded image.
<SiFuh>
farkuhar: They can't lift from shoulder height upwards. They are only strong enough to lift from waist to shoulder level. Hmm
<SiFuh>
Hands out in front. Not above.
<SiFuh>
farkuhar: Haha yeah, dlcusa's webp of Popeye
<SiFuh>
farkuhar: They all work in a big tyre shop nearby. Came over to help me lift the roof racks and tent off the roof. They said over 100 KG. I think only 70 to 80 KG.
<SiFuh>
Tomorrow, unbolt the roof racks and sand it down smooth and prime it.
<SiFuh>
2,000 MYR for custom roof racks. Good welder, shit at painting. So it is already rusting.
serpente has joined #crux-social
<farkuhar>
Here's another possibility: create only *one* overlay repo, and populate it with those ports that actually need patching. If the official CRUX ports build fine on CRUX-MUSL without any patching, we don't need to maintain a copy ourselves. That way we avoid the "sissy job" of pushing version bumps every day.
<farkuhar>
For the purposes of building the ISO, though, it might be necessary to merge the overlay into the official repos. Or the Makefile could be told to prioritize the overlay repo, when dups are found.
<SiFuh>
farkuhar: ukky: On the mounted ISO where is setup-chroot located?
<farkuhar>
SiFuh: you mean the live ISO environment? Check under /media/crux*
<SiFuh>
Hold on. I will boot ukky's ISO because it is still on a USB
<SiFuh>
It's in /usr/bin/
<farkuhar>
Yes, of course. And you find setup-helper there too?
<ukky>
farkuhar: okay, I will create 'muslcrux38' overlay that only contains modified ports, where Pkgfile and/or footprint was modified
<farkuhar>
ukky: Thanks
<SiFuh>
farkuhar: yes
<farkuhar>
The "one overlay" method is essentially what emmett1 uploaded on his codeberg instance. But emmett1 stopped updating his repo 7 months ago.
<dlcusa>
SiFuh, I saw that as well as the go-around Tuesday morning at Washington National. It's clear to me ATP's are more alert to ATC errors now.
<SiFuh>
Yes
<SiFuh>
farkuhar: I am stuck. If I run chroot /mnt on a system previously that had run setup-chroot then my script is fine.
<dlcusa>
First-year ATPs have not been given DCA routes for decades, home of the infamous "river approach".
<SiFuh>
If they hadn't run it and I run the script them my script is not. Hmm I have an idea.
<SiFuh>
It should check if you are have /mnt mounted, if it has already been chrooted then chroot and build the kernel. If not it will do the same as setup-chroot and then chroot and build the kernel.... hopefully.
<SiFuh>
farkuhar: So I should modify my first script so that if gcc doesn't exist on the ISO then it should error out and say no gcc found :-P
<ukky>
SiFuh: script is good. About .config file, maybe we can save it before 'make mrproper, then restore. This will help the user to base their 'make menuconfig' on the config used to build ISO.
<ukky>
Config from ISO is copied as /usr/src/linux-*/.config
<ukky>
That's wrong location
<SiFuh>
ukky: Ahh yes, I forgot that part
<SiFuh>
We can grab the config from contrib or another location
<ukky>
crux/kernel/.config or crux/config/contrib/config*-modular on the ISO
<SiFuh>
ukky: And I should comment out the lines export JOBS=$(nproc) like the pkgmk.conf I think.
<SiFuh>
ukky: which on linux-6.6.62.config or linux-6.6.62.defconfig ?
<SiFuh>
Isn't defconfig what make menuconfig will make anyway?
<ukky>
linux-6.6.62.config has jaeger's modification and what is being used to build kernel to boot ISO.
<farkuhar>
Someone suggests that /bin/mount commands should be protected behind a /bin/mountpoint test, and then they go even further, adding all these nested printf commands to inform the user what's happening.
<SiFuh>
I have not used VERBOSE and not sure why it is there.
<SiFuh>
The colour? Looks like bold white only.
<SiFuh>
So it should be USE_TONES=false
<ukky>
SiFuh: I did not find any obvious faults, I think it is perfect now.
<farkuhar>
I think I did see color when I tried USE_COLOR=true, but maybe it depends on the drivers that your kernel has for console graphics.
<SiFuh>
ukky: I haven't run it. So it is untested
<ukky>
SiFuh: Your brain and mine already ran it a few times, so it is semi-tested.
<SiFuh>
I've been working it longer than a few times. :-P
<farkuhar>
SiFuh is referring to kernel-building in general, not this script in particular?
<ukky>
Okay, ran through SiFuh' brain 27 times
<ukky>
farkuhar: just this script
<farkuhar>
No matter how many times I run dlcusa's base64-encoded text stream through my brain, I can't implement the decoding algorithms fast enough to see the Popeye image.
<SiFuh>
farkuhar: This kernel in chroot script. Because I don't like mistakes and ukky found a big mistake in mine
<SiFuh>
farkuhar: I missed the part where I whacked the .config if it existed and never pushed a .config in to the source.
<SiFuh>
farkuhar: That means the non-chroot build is missing the .config too
<ukky>
SiFuh: just a note, but I never build kernel in-place. I always create 'build' directory outside of kernel sources. All object files and binaries will be there.
<farkuhar>
SiFuh: The handbook will still list all the steps that your script is automating, just as it does when introducing the `setup-chroot` script. You shouldn't worry about users not being able to recover from a missing config file.
<SiFuh>
farkuhar: I just hate typing it. ;-) Hence creating ckut so it can just look online, grab the kernel I want and do everything with little input from me
<ukky>
SiFuh: you would, if you start cross-compiling on the same host from the same kernel sources.
<SiFuh>
ukky: Ever tried my ckut script?
<ukky>
SiFuh: no
<ukky>
What does it do?
<farkuhar>
SiFuh: Outside of chroot there's no gcc available, at least on the live ISO. Are we back to proposing that a working gcc be included in the live ISO environment?
<SiFuh>
ukky: It looks just like crux setup script. It checks online and lists available kernels and you can select one. Or you can manually input what version you want. It can download it, extract it. Prepare it. Run make menuconfig. The build and install it. There are options to choose editor and locations in the /etc/ckut.conf file and auto detect if you have lynx installed so you can go to the linux kernel
<SiFuh>
site. And auto detect which bootloaders you are running and some options for the bootloaders although, not much for grub since I don't use it.
<SiFuh>
Still a work in progress. I want to slim the code down a bit
<SiFuh>
Also ukky I'd like to re-write in sh and not bash.
<SiFuh>
Or maybe in ansi c if possible
<ukky>
SiFuh: It looks like a good script. I just use a little helper script to use config from previous kernel in a new kernel build.
<farkuhar>
ukky: Isn't there a Makefile target "oldconfig" for that?
<ukky>
farkuhar: there is. My helper script takes old config from storage and copies it to a new build directory. Plus, creates some symlinks to point to latest build directory.
<farkuhar>
Okay, so just the usual workarounds when you insist on out-of-tree builds.
<ukky>
farkuhar: is there a generic method how to distinguish port sources (downloaded, external) from local port files (patches, etc)? If I delete all *.{gz,xz,bzip2}, some patches might be deleted too.
<farkuhar>
ukky: prtwash and prtsweep take two different approaches to that task. One of them uses .signature, the other one sources Pkgfile and echoes the entries in ${source[@]}. I forget which method goes with which tool, though.
<farkuhar>
ukky: the git repo should still have the bash counterparts of prtwash and prtsweep, if you just want something to copy (rather than translating from Perl to bash).
<ukky>
farkuhar: thanks
<SiFuh>
farkuhar: I was writing it in ansi c by the way for fun. I got stuck on a few things. I will look into later another day. But I do like ansi C over sh.
<SiFuh>
farkuhar: I have everything (check /mnt is mounted, check if chroot exists, if not setup chroot, even chroot works) up to the part where it copies the config over working. And just barely touched the kernel build process.
<SiFuh>
Oh and uses a setup-kernel.conf file. I think I will add a flag in it to make sure the user edits to configuration file first before the program proceeds
<SiFuh>
ukky: They actually sell boxes pre-made that I can use those batteries I have for a charger. They are super cheap too. I will buy and try one. I don't like it, I can modify the case and change the circuit to the one I have.
<ukky>
SiFuh: that's a good option instead of building your own case.
<SiFuh>
ukky: Yeah considering the price. And there are a few out there some a little more expensive. But worth a test.
<SiFuh>
And I can always poach parts ;-)
<SiFuh>
Last time I dropped my truck off at the mechanic, they fucked around with a black box under my dash. They should never have touched it. But I guess they were curious. I have a fused junction box with two bus bars that all the accessories are connected too. So when I got home, I couldn't open my gate. Checked the box and some of the wires were loose. :-) I have a key fob attached to my headliner that is
<SiFuh>
wired to the junction box because it uses a 12 volt battery.
<SiFuh>
I hate fobs. I don't want to carry that crap around or constantly changing the battery. So having it wired direct to the truck, solves all my problems.
<SiFuh>
ukky: Sometimes I deliberatly lock my keys in the truck when I am in the jungle and going on walks. But before I do I change my antenna for the UHF/VHF radio. I can use that antenna to push the unlock button up on the truck door. ;-)
<ukky>
Key fob in waterproof container would be safer way to unlock your truck
<SiFuh>
My truck doesn't use a key fob. It's old school 1990's it needs a key
<SiFuh>
ukky: I use the fob for open the gate
<ukky>
You can put key on the top of tire, like in the movies
<SiFuh>
We do that when we drive Taxis.
<SiFuh>
I have a nice place to hid the key though.
<SiFuh>
I should take a photo
<ukky>
I would just keep a key on some necklace (not jewelry)
<SiFuh>
You can't see it unless you look inside of it. Because it goes aways back.
<SiFuh>
But if I am more than 15 minutes from the truck, I locked the keys inside the car
<SiFuh>
ukky: Hahaha the stupid dog. Good thing it is censored by the garbage bag strap.
<ukky>
I noticed dog too. My first thought: it is smoked chicken
<SiFuh>
ukky: Wife keeps giving her a bath. After, she gets a skin infection. I told the wife. Stop bathing her. She is a dog. 6 months is fine. She says "But she stinks" I said "Stop smelling her then"
<SiFuh>
ukky: My brother washes his dog every night. But he refuses to listen to any of us. Fortunately the dog is super clean and has no issues at all.
<SiFuh>
farkuhar: I don't know why Tim has to reply to people saying that he doesn't use something. It's pointless.
<zorz>
they were speaking in #voidlinux for minisforum the other day... first time i heard of them.
<SiFuh>
Probably Taiwan or same company as mine
<zorz>
what i am concerned more is the fun.... it is on top of the cpu... looks good.
<zorz>
can find it from amazon.de its like shoping in greece.
<zorz>
fan
<zorz>
hahaha fun fan
<SiFuh>
I was thinking because i have two, I might link them together and run one OS on two machines to double it.
<SiFuh>
But lazy at the moment because I got many other projects to do
<zorz>
they are good for monero aswell
<SiFuh>
I used paypal for mine they suspended my order and asked me for a photo of my CC
<SiFuh>
I told them to get fucked because I used Paypal and not a CC
<SiFuh>
Then they approved my order and sent it to me
<zorz>
since you pay, what is theur problem ?
<SiFuh>
Thank you very much for your cooperation! We need to confirm whether the order is real and valid, and our email will also be used as the delivery certificate. Please confirm the order information. If there is no problem, we will arrange the delivery for you now.
<SiFuh>
Have you recently purchased a MiniPC from MOREFINE (www.morefine.com)?
<SiFuh>
As we have seen your order in the system back office and it is flagged as high risk,
<SiFuh>
We will need to carry out a series of risk assessments first. pls offer the Front and back photos of the payment credit card and a screenshot of the PAYPAL payment.
<SiFuh>
So I replied
<SiFuh>
No, I will not. That is ridiculous. Not only is this against all banks regulations, it is absolutely dangerous. Under the banking regulations only the last 4 digits are to be shown or stored in a database. Besides, I paid with paypal. My first transaction didn't go through. The second did. I also phoned the bank that night which is located in Australia to inform them that this transaction and another
<SiFuh>
on Ebay was made by me. The lady said that both transactions are fine and there were no issues. What I will give you is the line from my bank statement and my paypal receipt. If that is not acceptable then I am happy to receive a refund.
<zorz>
i am going to amsterdam.... and booked the ticket with ryanair. usualy when i pay i get a notification in mobile and i accept. This time they ask me to pass full credentials. and then had the notification to approve in my mobile device
<SiFuh>
Ryan air. HAHAHAHAHAHA you idiot
<zorz>
80 euro
<zorz>
flying the old good boeings
<zorz>
hahahhaa
<SiFuh>
I remember I booked a ticket on Malaysian Airlines. When I went to collect the ticket they asked me for my credit card number. So I gave them my medicare card. My Malaysian friend was pissing himself laughing.
<SiFuh>
He's one of the few comedians that can make a crowd, just laugh and laugh and laugh over and over again. No breaks. And it is all 100% original stuff he comes up with.
<SiFuh>
And you can still watch it and laugh again, even though you expect it.
<SiFuh>
He has some classics. The electric toothbrush. Great Western Railway. A potato.
<SiFuh>
Haha, it is 4:30 am here. I can't see hahaha
<SiFuh>
ukky: Im still tossing around the ide of taking the wife to Australia to meet my parents
<SiFuh>
But I want 3 weeks not 7 or 8 days
<ukky>
Yeah, 3 weeks is a minimum for traveling. 7 days is not enough.
<SiFuh>
ukky: No, for me minimum is 52 weeks
<ukky>
Ha
<SiFuh>
And get a job
<SiFuh>
Yes, that is how I work
<SiFuh>
I hate it when people say they know Thailand and they took a tourist trip for 14 days.
<SiFuh>
I did 8 years. I know Thailand. I travelled 3rd class, learned the language. Travelled with friends and alone.
<SiFuh>
China was 4 years
<SiFuh>
Kyrgyzstan was 2 years
<SiFuh>
Every other country was 3 months.
<SiFuh>
Hapan was 8 days. Myanmar, Laos, Cambodia was less than 4 days.
<SiFuh>
Hapan = Japan was 8 days.
<SiFuh>
Actually I was pissed off Japan was 8 days. I had more I wanted to do. So I am going to have to come back. But next time. I am going to the mountains and forests
<SiFuh>
None of this shitty city crap and tourist garbage
<SiFuh>
ukky: Besides if I go to Australia, I will renew my passport there.
<SiFuh>
US passport
<SiFuh>
14 days
<ukky>
If your australian passwort has expired, how would you go to Australia? Do you still have US passport valid?
<SiFuh>
Reverse. AU passport is fine. US passport is expired. But I didn't want renew under Biden. Can't stand that cunt.
<SiFuh>
And if Camal Toe Hairy Arse got it. I'd never renew it.
<SiFuh>
And as much as I dislike Trump, I prefer him over those two fscking retarded imbreds
<ukky>
But you can still go to US and cover your eyes when you get close to Biden or Harris.
<SiFuh>
No, passport expires after 5 years after expiration date. Big process to re-apply
<SiFuh>
And the US would be fucked under her.
<SiFuh>
No point going back.
<SiFuh>
ukky: I like my scripts perfect. I like everything perfect. They are blemmish. I don't a passport under a blemmish.
<SiFuh>
Heh missed some words
<ukky>
Still, if you want easy passport renewal procedure, it is better to do it before it expires
<SiFuh>
AU 2 years after expire. US 5 years. Process is still the same as if not expired
<SiFuh>
Mine expired just before Resident Bidet got in. So no chance to renew
<SiFuh>
And I didn't want to carry a book during his era
<SiFuh>
His destruction was 4 years. The book lasts 10
<SiFuh>
ukky: I actually can't believe there are pro Bidet, pro Hairy Arse and pro democrat still.
<SiFuh>
It's like the are freaking blind to what was actually going on.
<farkuhar>
ukky: Thanks, I downloaded your muslcrux38 repo. I'm curious to see libxcrypt in there, though. I thought you didn't build that package for the ISO.
<SiFuh>
The democrat part was hijacked by anti-americans
<SiFuh>
farkuhar: It isn't in the ISO I checked
<SiFuh>
farkuhar: It was pointless to even bother with building it since musl has its own.
<SiFuh>
And not to mention libxcrypt is basically a crypto version of SYSTEMD
<SiFuh>
So maybe zorz is correct. Beerman will eventually push SYSTEMD
<farkuhar>
I'm also curious to see sudo.pam still included with the sudo port, despite the sudo Pkgfile having ./configure --without-pam --without-pam-login
<ukky>
farkuhar: I think it was much later when I realized that libxcrypt is only required for glibc
<SiFuh>
farkuhar: I searched for pam shit and didn't find anything. I will check again tomorrow between sprays.
<ukky>
muslcrux38 repo contains all of changes I made, relevant or not.
<SiFuh>
ukky: I told you like day one dude
<farkuhar>
Oh, and openssh can also have its sshd.pam removed, as long as we're tidying up.
<SiFuh>
So basically farkuhar what you are saying is ukky didn't stip PAM from everything?
<SiFuh>
grep -r /usr/ports {PAM, pam}
<SiFuh>
Does that even work? I did something like that to strip PAM from everything.
<SiFuh>
ukky: I think I will build a version off of yours without pam/PAM and change core to / with stenur's overlay or my own script and opt into /usr/
<ukky>
/etc/pam.d/ is not present on a full install.
<SiFuh>
ukky: I didn't see it too. It was the second thing I checked
<SiFuh>
ukky: Maybe need a nopam-core nopam-opt and whatever
<farkuhar>
All the `prt-get isinst` tests return false during stage0. What about stage1 and beyond?
<SiFuh>
stage 1 is not stage 0
<SiFuh>
Stage 0 builds from your current system
<SiFuh>
Stage 1 builds from you stage 0 ports in a fakeroot
<ukky>
prt-get isinst returns false in stage 1 and 2 for all queries, because /usr/ports directory is empty.
<SiFuh>
Huh?
<farkuhar>
Thanks for explaining. So it's stage 1 and 2 where the `prt-get isinst` tests all return false.
<SiFuh>
I need to check that ukky and farkuhar
<SiFuh>
But as far as I know once stage 0 extracts for stage one isinst work but in the fake root.
<ukky>
Results in stage 0 are also pointless, because those are results for host packages
<SiFuh>
When I built the Beelink custom ROM for CRUX I remember having to add ports I didn't really need in Stage 1 to keep Stage 1 happy
<farkuhar>
Now the only difference between opt/geeqie and muslcrux38/geeqie is the removal of a line that tests for the presence of bash-completion. I think that line is harmless, if you're building geeqie in stage 1 or 2, and the target system has no bash-completion installed.
<SiFuh>
farkuhar: Here is an idea. Why not abolish the shit. And you build it later?
<ukky>
farkuhar: opt/geeqie had 'git merge' leftovers, that's the reason I edited Pkgfile
<SiFuh>
We can create a ports -u version for crux-musl but for the ISO why bother? Slim it down.
<SiFuh>
You built the kernel, can surely build a port
<SiFuh>
No need out of the box
<SiFuh>
ukky: farkuhar: Let's say we strip down the ISO to core and some opt packages. Get's the system up and running. But after a ports -u. The patched versions are available for compilation. So they can add their stuff later?
<farkuhar>
This is ukky's ISO; let's appreciate all the work that went into it, bringing it as close as possible to the official CRUX ISO. There's no need to try steering us back to a core-only ISO.
<SiFuh>
farkuhar: I do, and don't forget I also built every port too. Except three and went camping :-P
<SiFuh>
I've working ISOs here but I won't release them. I am happy with ukky's version so far.
<ukky>
SiFuh even went further and asked if I will continue the work on musl ISO so he can take a break
<SiFuh>
But as I said. Lets say we strip the ISO down. Then the user can build upon it
<SiFuh>
ukky: I did 3.7. That is where it was suppose to end. But that asshole dude here was pushing me to produce a 3.8
<ukky>
Do I know him? :)
<SiFuh>
ukky: My goal was to create a base for 3.7 in MUSL that we can build upon. Users want to try. No cross comping.
serpente has quit [Remote host closed the connection]
<SiFuh>
ukky: Yeah his name starts with F
serpente has joined #crux-social
<farkuhar>
Good thing serpente missed that remark.
<SiFuh>
He didn't
<SiFuh>
He is a close friend so :-P to you
<farkuhar>
The explanation directed at ukky is what got missed, not the original "asshole dude" reference.
<SiFuh>
besides, I already sent it to his EMACS
<SiFuh>
So he misses Jack Shit ;-)
<SiFuh>
Okay the original plan was to help those who can't cross compile. I did that. Then I called on emmett1 for assistance so both he and I can produce the core system so it can be built upong
<SiFuh>
So 3.7 came out with a core only ISO
<SiFuh>
That is where I was done
<SiFuh>
The "asshole dude" came in and wanted 3.8, so I started it and completed most of it. Then ukky stepped in and produced something good too.
<farkuhar>
joe9 in #crux said it's been 10 years since he last created ports. I could have sworn I recall a joe9 repo in the portdb long ago, but it's no longer listed. Maybe I'm remembering the portdb as it stood in 2015?
<SiFuh>
Both he and I were working on something closer to the offical CRUX ISO
<SiFuh>
But I had to take time off. ukky took over and as I said. He produce a nice ISO
<SiFuh>
He did some things different from me but still acceptable.
<SiFuh>
My goal is to push an ISO that follows the exact same build process as jaeger
<SiFuh>
This is where emmett1 and I had a falling out so to say... He was cutting corners. I didn't want that. But he assisted me even though I abused and bitched to him.
<SiFuh>
To be official we need to follow the official way even if _you_ don't understand why.
<farkuhar>
So beerman throws abuse at a newbie, and that's crossing a line. But abuse and bitching directed at emmett1 is somehow acceptable?
<SiFuh>
I can use the ukky base system, but I think I will run into the same error process. I don't know why maybe ukky knows why. Why can't CRUX Musl bootstrap successfully?
<SiFuh>
emmett1 is not a newbie and he is very skilled. And I love that guy. He and I do many things together. He was taking short cuts. And I couldn't accept it
<SiFuh>
CRUX way or no way. Simple.
<SiFuh>
And if you didn't notice CRUX is CORE nothing else. That is why I get offended by linux-pam and sudo and dumb_runtime_dir in the core ports
<ukky>
SiFuh: all 'make' stages (0, 1, 2) were successful for my build.
<SiFuh>
ukky: Nice...
<SiFuh>
But I think I will push for a core ISO. Haven't decided yet.
<ukky>
All stages were built on glibc system in chroot environment
<SiFuh>
Yeah, nice, but I prefer to buil it inside a MUSL system,
<SiFuh>
ukky: I built my last one inside of a glibc system just for laughs
<ukky>
My other systems are not powerful enough
<SiFuh>
ukky: You said don't comment out nproc shit. I had to. My system is powerful enough, but it over heats and shuts down.
<SiFuh>
So I persuaded myself to comment it out for those who don't know what is happening
<ukky>
Oh, then it is okay to comment-out. The user can uncomment if needed.
<SiFuh>
That's my thinking. Same as pkgmk.conf
<ukky>
current (3.8) git port also requires a patch for parallel build, otherwise it fails
<SiFuh>
ukky: I still can't believed I missed the .config part.
<SiFuh>
Thank you for pointing it out.
<SiFuh>
Such a dumb mistake
<ukky>
it happens
<SiFuh>
When I wrote ckut, I played every possible outcome in my head before I wrote the script
<SiFuh>
And I missed one important part for the setup-kernel scrip bahhhh ti sut!
<SiFuh>
That's Thai for most crazy in the world.
<SiFuh>
Even stenur was shocked. He said something like this "WTF man, I didn't expect this. It covers everything" in his broken English.
<SiFuh>
And here is old man SiFuh and he forgets the .config file
<SiFuh>
Now all I need is someone to test the script
<SiFuh>
ukky: I tried to call setup-chroot but it wasn't carrying the variables accross. So I ended up just inserting it into my script
<ukky>
Not me, I build kernel using commands in bash history. Plus, always as non-root, and my kernel sources are owned by root, so user cannot mess that up.
<SiFuh>
ukky: One user complained in #crux complained to me saying he builds kernel as non-root. I thought about adding that option then thought. I don't freaking care.
<SiFuh>
So I never added a non-root option. Extracting as root to /usr/src then changing the user and compiling, then switching back to root. Seemed frugile at best. So I just said... No to myself.
<SiFuh>
ukky: Actually I do have a plan to switch ckut to ncruses and ansi C. I just haven't the time.
<SiFuh>
Besides. Dialog is great. So if I stick with dialog, I will need to slim my script down as I do plan in the future
<SiFuh>
And convert it to SH and not use BASH
<ukky>
SiFuh: I think it would be easier if you remove chroot support and copy your script to /mnt/usr/src and then user can manually chroot and build the kernel with your script.
<SiFuh>
Yes, I thought of that ukky but newbies?
<SiFuh>
Actually pieces of my script can be used in jaegers chroot script
<ukky>
Even newbies need to chroot to setup the system in Crux
<SiFuh>
Try running setup-chroot twice and watch the mess unfold
<SiFuh>
Actually I might ask jaeger about that
<farkuhar>
Maybe also ask jaeger for the log of the stage 1 or stage 2 xorg/mesa build, to see whether the `prt-get isinst` tests all return false.
<ukky>
For newbies there is Ubuntu. Like that podcaster, why would someone attempt to fly an aircraft when they cannot master the ride on bicycle?
<SiFuh>
Done!
<SiFuh>
farkuhar: Modify the Makefile like everyone else does.
<ukky>
farkuhar: I will find the log from stage 2
<SiFuh>
ukky: My stage 2 kills it.
<ukky>
SiFuh: stage 2 kills mesa build?
<SiFuh>
Stage 2 kills many things ukky
<SiFuh>
Stage 0 success. Stage 1 some things break and I fix it. Then stage 2 several things won't compile
<ukky>
SiFuh: Try updating you gcc with patches and options from my port
<ukky>
s/you/your/
<SiFuh>
ukky: Eventually. Not now. But I bet we still release MUSL before CRUX 3.8
<SiFuh>
ukky: Done, jaeger will look into it
<ukky>
farkuhar: https://0x0.st/8AFL.txt <=== This is how all those prt-get tests look like when building ISO
<SiFuh>
farkuhar: I like jaeger. I have known him longer than any of the other CRUX users.
<SiFuh>
ukky: Hmm I don't have that problem [Config error: can't access /usr/ports/core]
<ukky>
SiFuh: Do you have stage 2 build log?
<SiFuh>
Yeah on an NTFS external drive
<farkuhar>
ukky: It looks like stdout and stderr all went to the same place in that output. You eventually get the desired info from `prt-get isinst`, but not until after the noise about config errors.
<SiFuh>
I can dpaste the top section tonight sometime. It is morning here and I am quite lazy
<ukky>
When ISO is built in stages 1 and 2, ports are located at /src/ports in chroot environment, but prt-get looks only at /usr/ports
<SiFuh>
Hmmm
<SiFuh>
I will check in the evening
<SiFuh>
Maybe it reuires a git/rysnc if I am wrong inside of the fakeroot
<farkuhar>
That can be configured, can it not? Just change the "prtdir" directives in prt-get.conf, or make an alias, prt-get='prt-get --no-std-config'
<farkuhar>
alias has to be placed in the right place, though... somewhere that will affect every pkgmk process, i.e., /etc/pkgmk.conf
<ukky>
farkuhar: Yes, it can be configured. ISO Makefile has to override /etc/prt-get.conf.
* SiFuh
is afk to keep farkuhar happy
<zorz>
you wrote his handbook ?
<zorz>
;pP
<ukky>
farkuhar: It is not done yet, I just noticed those failing queries when I've built musl ISO. But this is probably a big issue.
<farkuhar>
ukky: No worries. We've done a lot of testing today, learning a few things for the next ISO build. Nothing major that would block a timely release, though.