all of those crc/checksum generating apps can be installed that isnt really a pray it runs/compiles on BSD situation

Let's face it. BSD was born different from other systems.

Even back in the day BSD had a non-standard form of df. A non-standard cron. A non-standard lpr. A non-standard file system group inheritance feature. A non-standard uucp system. Restartable signals. Lots of things have been done differently on BSD since the very beginning.

It's always been necessary to know both sides of the fence when one is straddling it.

The one that annoys me the most is sed's -i in place edit option. Perl did it first and everyone else copied the interface. Everyone except BSD which did it differently. Aarrgh!!! That one hurts me the most.

i am new to bsd and im messing around.

i am playing in a vm

Welcome to the community! And have fun messing around.

because im a maniac i decided i wanted to use ports as opposed to pkg

but i couldnt figure out how to use git to manage the ports tree,

I think you should recompile the kernel, base, and ports and have a fully self-hosted and self-compiled system. FTW!

so what ive done is a fresh install

*not* include ports

install git with pkg

clone the ports tree to /usr/ports

pkg remove git

pkg autoremove

and now im installing git from ports

is this sane?

was there a better way?

i dunno...

Insane? No. Better way? Maybe. I would have done a git clone on a different system. Then made a ports.tar.gz of it and copied it over.

Then there would be no need to bootstrap git through pkg.

heh

i see exactly where you are coming from

but that would have required TWO vms

i guess the next step will be figuring out how zfs actually works

Okay. Your first mistake is that you are not running a free operating system on your laptop! :-)

ofc not

im using macos on my laptop

:P

macos on the laptop

junos on the switches

Then you could have done a git clone and tar on your Mac. That should have worked okay.

why else would i be interested in freebsd

yeah that makes sense

shit i should have thought of that

just clone the repo anywhere else

IRC has to be good for something. Like reminding people of obvious things. :-)

and then copy the whole damn thing to the freebsd machine

Yes. Now you have it!

you could also download a tarball from freebsd.org

its weird though

i found this post: forums.freebsd.org/threads/guide-us…orts-source-and-documentation.79721

it seems to touch on the problems

kevans way is another great way to bootstrap. However then that tar bundle I don't think includes a .git directory and so would require another git clone at some point in the future.

and even hint at a solution but it doesnt make sense to me what he is saying

he keeps leaping from /usr/ports (which seems very official)

to /home/peter/temp/myports

i just got confused and decided to do my "install git with pkg and then 'cheat' trick"

i guess my strategy should be fine

FreeBSD has migrated version control systems a few times already. And each time it takes people a while to get comfortable with the new tool. Which means that there is a lot of blogs and articles written about how to do it. But in the end it's not about the version control but about the build system which builds the code.

yeah thats what i thought

yeah, no .git. I'd love to write something to unpack a .bundle and get the project to officially distribute bundles

its slightly "in flux" and i should just chill out and try and learn

Traditionally FreeBSD builds things in /usr/src for base and /usr/ports for ports and the build is done as root. Which just feels very wrong when you are used to working in local sandbox working copies as non-root. But you can build in a jail which makes it about the same.

Think about your VM for example. You just created it. So if you compiled as root and for some reason the build did something terrible as root and damaged the system, well, now bad could it be? It's a freshly constructed VM. Just build another VM. It's like that when building in a Jail.

the thing i find *mildly* frustrating and im cutting some slack because it is a transition (or seems to be)

is that the docs say you can cd /usr/ports/devel/git make install clean

and then clone the git repo

but that WONT work if youve installed ports from the installer

and there is no "official" obvious fix for that

but im putting that down to a teething issue

well i say thats what the docs say,

thats how i parse it

and im an idiot

so i possibly should be less judgemental

but its cool

thats why im playing in a vm

There is probably no official docs on how to work through that part of the bootstrap. One is left to understand things and work through it as a homework exercise.

indeed

and i guess that maniacs that want to build most stuff from source in ports are treated as an edge case any3way

which is equally reaonable

And after you get past that initial bootstrap then everything is in steady state and can go for years without needing to worry about it again.

And it is easy to copy over to other machines when provisioning them.

Things also evolve over time. That's the traditional way to build. But the current way that all of the cool kids are doing is to run a pkg builder poudriere to build pkgs from source and then we install the packages on the systems. That's "new" in the grand scheme of things. But it's the better way to go for ports/pkgs now.

And it is still experimental but the base system is going the way of pkgs too with a PkgBase migration that is not yet in a release state.

ive messed with portmaster a bit

how does poudriere fit in with that?

It's a very good learning experience to know how to build ports from source though. Because that's how compiled binary pkgs are created. And that sometimes needs to be debugged and extended. So knowing how to compile ports helps with running poudriere building pkgs.

is it like a replacement / alternative to portmaster?

yeah its interesting: im a linux guy predominantly and while i favor debian atm, i used to love gentoo

I'll just say yes. But I don't think it is a completely binary answer.

its really intersting comparing gentoo to freebsd

(well if you are using ports anyway)

poudriere is similar-ish in spirit, but it does clean builds and it's geared more towards building package repositories in general rather than something you're intending to install locally (though you can do that)

I've come back to BSD after many years with Debian, and more recently Devuan, myself.

synth is a closer analog to portmaster that still does clean builds

i have a n900 that runs maemo which is based on devuan

I can't believe I was having glitching and artifacting on an RX 5700 with quarterly drm-61-kmod that seems to be resolved by the "latest" version of it

I'm mixed... I'm glad it seems resolved, but disappointed that a card this old was still recently having issues

Hi guys, why am I getting this: repository FreeBSD contains packages for wrong OS version: FreeBSD:13:amd64 # uname -a

FreeBSD nyc 13.1-RELEASE-p3 FreeBSD 13.1-RELEASE-p3 GENERIC amd64

yashi: what does `readelf -n /usr/bin/uname` say?

Notes at offset 0x000002c0 with length 0x00000048:

Owner Data size Description

FreeBSD 0x00000004 NT_FREEBSD_ABI_TAG

ABI tag: 1301000

FreeBSD 0x00000004 NT_FREEBSD_FEATURE_CTL

Features:

FreeBSD 0x00000004 NT_FREEBSD_NOINIT_TAG

description data: 00 00 00 00

yashi, Your uname -a says 13.1-RELEASE-p3 and 13.1 is out of support. The repository has moved on to 13.3. You will need to upgrade your base system to at least 13.3 in order to keep using the repository.

13.3 is very soon to be out of support so you should plan to move to 13.4 as well.

Anyone online?

yes

people in china

china moment

any ath0 users awake? after hearing all the good support it has, i got me one... and its been a nightmare....

constant ath0: device timeout, and this is about 12m los from the AP

my "iwlwifi0: Detected Killer(R) Wi-Fi 6E AX1675x 160MHz Wireless Network Adapter (210NGW)"

works so much better, but limited in speeds...

on fbsd that is, it rocks in Win 11

wettoast: I'm not 100% sure about FreeBSD support, but the Intel AX210 has been good for us on Linux

is there a way to fetch a non-github git repo with submodules in a port?

hi

If I've never set up a firewall and I want to start using fail2ban on 14.1-RELEASE, is it safe to assume I should do the ipfw version?

yashi: you use either ipfw or pf or ipf - all will work

yashi you can use PF/IPFW. No problem.

ok thanks

devnull: are you stalking me?

i just installed fail2ban

can someone try to hack me?

<_<

spoonix2ipfw: i see

wonder what chipset my workstattion will have, its and asus Zen 5 board w/ WIFI 7

any C guys know what this means? mbuf.h hasnt changed in a while...

if_re.c:1270:37: error: incomplete definition of type 'struct ifnet'

1270 |         sc->re_rx_desc_buf_sz = (ifp->if_mtu > ETHERMTU) ? ifp->if_mtu: ETHERMTU;

|                                  ~~~^

/usr/src/sys/sys/mbuf.h:139:9: note: forward declaration of 'struct ifnet'

139 |         struct ifnet *ifp;              /* network interface tag belongs to */

probaly you missed to include the header including the struct ifnet declaration

look at ifnet(9) to find the correct includes

llvm problem? if_re.c:4623:43: error: call to undeclared function 'IFP2ENADDR'; ISO C99 and later do not support implicit function declarations [-Werror,-Wimplicit-function-declaration]

its not my code, but we did get llvm19 recently

im no dev tho, ill pass that on to the port maint though, ty.

which port?

net/realtek-re-kmod/

doesn't sound like a llvm problem, more like some details in the include chain have changed and upstream hasn't updated the includes

upstream of the port right?

yes

got it, ok ty

something like 'struct ifnet' was somehow implicit included by some header, but isn't anymore

cool, i'll let the maintainer know

satanist has it exactly. The "incomplete definition of type" with a struct means that the compiler saw a struct declaration and is now seeing a struct use but has not seen the struct definition (the details, the struct foo { int j,k,l; }; part). That's valid when dealing with struct pointers but not for struct content access.

so my brain says.... some guy at Realtek made made a boo boo?

havent done C since 20 years... lol

how should freebsd-update.conf looks like for binary update. is ode set: 0

is Components src world kernel enough, please?

nsoci: i always compile, but this should work, also see its man page

thank you

thats for main branch btw

man freebsd-update.conf

tooo

will do

also, check this out

YMMV

Good effort.

Is csh still the default shell for the root user? Tried to change the shell for toor and messed up, now root has fish shell, it works but does not feel right, better to have the default for root ;-)

I actually have bash for root right now, and am debating doing fish for my main user

I can actually just use profiles in konsole to accomplish the user part though, don't actually need to "chsh" it for them

thorre: default shell for root has changed to /bin/sh

satanist: thank you

bash as shell for default can backfire when something goes wrong with shared libs in packages.

that's why bash-static exists

yes

But it only helps when it is the one being used.

I used to build a static bash and install it to /bin/bash, so I could change root's shell.

If you're using the bash port in the default location, changing root's shell can bite you in the ass, if /usr (or /usr/local) is a separate volume.

These days, I jsut have root's shell has default and 'sudo -s' when I need to be root, which keeps my own shell.

By default FreeBSD has both root and a "toor" template. My choice is to leave root stock with /bin/sh so I can always log in but then I set up toor as a root user with bash. Then if bash gets broken (I broke it on an update once) and I can't log in as rwp using bash then I can always log in as root. And toor is also available. "sudo -i -u toor" I use a lot.

I thought about doing it the other way around with toor using /bin/sh but then I would not be using toor and would forget about it through lack of use. So I force myself to use it all of the time.

,/1

(woops)

whois

should be asking who is *not*, these days

I don't think therefore I am not

you wish