reporting, claws-mail/sylpheed no longer support outlook, given that libetpan is not built with sasl support

CCFL_Man: you get it going?

Howdy, folks!

Well, i managed to test out running HL1 via LSU, and both SP, and MP over the internet works pretty darn well.

I'm still trying to get OpenGL on linuxulator to use amdgpu, to no avail. So HL1 crashes on startup when using OpenGL, but software mode works!

Even though, i tested online MP via an empty server. But i plan to test it with a populated server.

Also, i tested UT99 on wine, it doesn't boot at all. But UnrealED works!

And for the record, this is without the oldunreal patch applied. So it's a 100% clean install.

nicholaus04: you probably get a few points for effort. But I would say if you want to play most games, and avoid Microsoft. Linux is likely your best bet

And i also managed to test out luac, and even managed to see how it would do if i intentionally caused an error in the script.

AKA: I discovered that luac hardly gives 2 cruds about weather or not an error was in the script.

luac?

nmz: The lua bytecode compiler.

of course it doesn't, syntax errors are checked on the lua compiler

; luac hello.lua

luac: hello.lua:4: ')' expected (to close '(' at line 2) near 'print'

nicholaus04: maybe its an old luac?

but once its compiled, lua doesn't care I think

nmz: Just using luac53.

Though, the lua script used for the compiled just has a print statement, and a undeclared function being called after it.

bytecode is version dependant so you can't use another, now that I think about it, my linter is actshually luac -p (I haven't touched lua in so long I forgot ha)

so it should've noted the syntax error

nicholaus04: do you mean you used 2 calls one after the other?

print()() ?

nmz: The script goes along the lines of: print('Hello, world!') duh(). (duh() isn't declared anywhere in the script.)

but that's legal lua

because _G.duh could exist

nmz: True.

but I do agree that lua's error checking is atrocious. however its dynamism prevents it from anything being done about it

And yes, i'm using a IRC client, in a vim terminal, in a VT(As in, no X.Org/Wayland).

yuck

its not that bad, the freebsd VT, is graphical, so you could have a VESA getty that even renders emojis

Oh and yeah, i'm installing HL2DM in the off chance i can somehow get amdgpu to do OpenGL in linuxulator.

I have no idea what any of those words mean

mostly HL2DM

nmz: Basically, i'm just seeing if goldsrc and source 1 based games can work on FreeBSD via LSU. Also, HL2DM = Half-Life 2: Deathmatch.

Though, i feel like if i did start HL2DM right now after it's done installing, it would crash on start up, since SDL2 couldn't create a OpenGL window.

So, i'll just wait for a solution to getting OpenGL to render on a amdgpu based GPU under linuxulator.

GL

I don't even have hardware acceleration

can't even watch a video

so

maybe I should switch to void linux or whatever

nmz: Try Half-Life in Xash3D-FWGS.

mount_fusefs: /dev/fuse on ___: Operation not permitted

got it

never been a hf person

redruM: why hf when quake exists?

Though, the short of the amdgpu problem on linuxulator for me, is that it doesn't seem to load the dri library for it, so it forces mesa to fallback to llvmpipe(AKA: the software renderer).

Even though, it's on there.

Though, there's no such library provided by any of the packages, but one of the other ones it tries to load (radeonsi) is present, but suffers the same fate.

nmz: "hf"?

nmz: Do you mean Half-Life, 'HL'?

ya

nmz: It's just an option for you, if you want it that is.

I think that suggestion should go to nicholaus04 no?

nmz: Correct, my mistake, I apologize.

Well, i got the whole "OpenGL on linuxulator" problem fixed.

Turns out, it was an X.Org config file goofing things over.

Though, it still complains about not finding amdgpu_dri.so. It now works pretty darn fine.

nicholaus04: Nice job in figuring it out.

apparently 14.1-p3 still has the realtek watchdog timeout bug

nicholaus04: I'm impressed that Linuxulator can go that far, my experience with it is limited to running Linux binaries of git-lfs, CMake, Pandoc, and such

nicholaus04: curious, which AMD GPU do you have and do you have any insight into how well is it working performance-wise compared to e.g. Linux?

afaik those have all real ports, so I'm curious why use linuxemu for them?

I accept your apology and absolve you of your mistake

whoever came up with it, thank you, I always have to rewrite this because checking to see if I'm online is for some reason, difficult

Ok obrońcom can run as user OpenVPN

But on 14 it requires an extra directive in the config

OpenVPN that is

Well, i tried to run HL2DM via LSU, and it bails out informing me that only one instance of the engine can be open. Only to find out it's failing to create a lock file for it for some reason.

Even then, i tried 2 stupid things to get it a little bit bypassed.

1 would be to chown the /tmp directory to be owned by my user, as opposed to root. No dice.

2nd would be to create the file that source is expecting, mostly to see if touching the file would trip it up on boot. Also nothing.

And i somehow managed to make it so steam can't start anymore.

It keeps trying to create some directories in /tmp, with all of them unknowingly failing.

Basically, it prints something along the lines of (DATE) /tmp/dumps(Or dumps01-09): failed to create, skipping/

Then a fatal error occurs thinking that theres still some /tmp/dumps directories still remaining, despite the fact that theres none.

According to kdump via ktrace: linux_access -1 errno -2 No such file or directory.

So somethings goofed up with /tmp.

Hmm...

Well, it's not the fact that /tmp isn't a tmpfs directory. (AKA: i tried unmounting /tmp, only for X.Org to stop working. Fortunatly, i have an entry in /etc/fstab so i can easily remount /tmp with a simple mount -al).

Unless it's referring to another /tmp directory, IDK how i can even fix it so steam can properly start up.

I somehow managed to fix it without help.

I just needed to symlink the systems /tmp to linuxulators root directory.

And with that, i was able to start HL2DM!

And yes, ingame performance is pretty darn good!

is there a way to allow a process to run chown on files/dirs without running it as root?

doas can be configured to only allow certain args, so you can give permission to run only on certain dirs/files.. i imagine sudo as well would allow for those types of restrictions

scoobybejesus: in my case though this is so a process ("syncthing") can do chown/lchown system calls on files so I don't think doas/sudo method you are giving will work as the process doesn't use doas/sudo.

docs.syncthing.net/v1.27.10/advanced/folder-sync-ownership.html, on linux for example they use setcap CHOWN and FOWNER capabilities on the process

for syncthing, i have it in a jail, and when i pull in files, my script will jexec into the jail and chown as root in the jail

maybe my way is clunky

hmm. what are you using the script for? In my case I run syncthing via ports directly in a jail

I was trying to avoid running it as root in the jail if possible

For some backup procedures, I just do zfs send/recv via sanoid. But I also have a script for backing up other certain files from my VPS or local box, and the script puts those dirs/files in a new folder named after the day.

syncthing is not running as root in the jail. the script is creating the dir in my syncthing/.backups/ dir, and placing files/dirs there. Once they are there, the script will jexec into the jail and chown the files/dirs as being owned by syncthing

maybe i'm oversimplifying, as i don't want to put a wall of text in IRC. but i'm letting syncthing just sync files. the issue is it won't sync files owned by root:wheel, so i chown them after placing them under syncthing's directories

So I have zvol for a vm, and I need to inspect the filesystem as its unbootable, as this is a block device... how would I mount it?

mount? why would you need to handle a zvol diffrent than a disk? what filesystem is on the zvol?

scoobybejesus: Got it, mostly. If syncthing is not syncing the files owned by root:wheel for you how are they showing up in the syncthing jail you have? I'm assuming you have a nullfs directory in your syncthing jail but that when the files show up there as root owned files, your script does the "chown" then.

yes, i have a separate dataset for syncthing, and i nullfs mount it inside the jail. from the host's perspective, i had to create a syncthing user with the same 983:983

I am wondering why syncthing is being run on a mix of root and non-root files... For me it is only ever used on my personal non-root files.

It's just not something I would consider doing.

polarian, You should be able to deal with the zvol exactly like it was a disk image. I imagine it has a GPT partition table and everything.

rwp: bsdlabel didn't work on it

nimaje: its gpt + FFS2 (OpenBSD)

but I can't see the disk labels, I get an error

just because i like certain things accessible on my laptop in the event of catastrophe or even convenience. maybe i want to check what the sanoid conf is or rc.conf... i am paranoid about losing data sometimes

rwp: I'm assuming you personally don't care or set any permissions on your personal files, hence why no need for running your syncthing process as root?

hm, not sure if freebsd can mount openbsds ffs2

I'm pretty sure they're the same.

adonis, My own files are created by my own processes and therefore owner:group:mode are all going to be okay as me by definition. No need for root in that case. I just run the process as me.

scoobybejesus, I guess for me those system things are backed up (Backup rule of 3-2-1) and I can always get to the backup separately. (shrug)

nimaje: a few people have told me that the UFS driver can work for most things

disklabel: error getting providername for /dev/zvol/zpool-storage/dns: Inappropriate ioctl for device

^^^ when trying to check the disk labels for the zvol

Has ZFS ever saved anyone any work?

CrtxReavr: tons

I haven't use any BSD before so I'm not very familiar with it, but I can't figure out why I can't install KDE. I added user level FreeBSD.conf file pointing to latest, but when I use `pkg install kde5` it says the package cannot be found in the repository.

As far as I can see, that's the correct name based on what I've found online. I also see it listed in Fresh Port. But when I use `pkg search kde` there's no kde5, or anything else that looks like it would be the KDE desktop. There are a lot of kde related packages though.

KMM: i think kde5 / kde6 is just a meta port to install a few kde things, such as plasma6-plasma, kate and konsole

anyone having problems connecting vnc to vm-bhyve on 14.1 ? updated yesterday and vnc hangs

rtprio: That's my understanding too, but it also says that plasma5-plamsa and plasma6-plasma are not in the repository.

But konsole is there.

KMM: the kde5 package (and many others) got skipped in the last build, as a transitive dependency (x265) failed to build pkg-status.freebsd.org/beefy20/buil…0amd64-quarterly&build=d032848c5848

nimaje: oh, thanks

nimaje: Thanks. Now at least I know I'm not crazy. When that happens you just can't install something?

possibly, in theory, should be able to build it yourself with the ports collection

KMM, Just recently we had a release upgrade to 14.1-RELEASE-p3 and that would have triggered the pkg autobuild of everything. KDE is large. It might not be built yet.

KMM, I am not a KDE user but right now in 14.1R I am seeing kde6-devel and it looks like I can install it okay there.

I don't remember seeings kde6 in `pkg search kde`, but I was scanning for 5, so maybe I missed it. I can't look now, but I'll check again later.

"pkg install -n kde6-devel" here on 14.1R shows "Number of packages to be installed: 342" but appears as if it would be successful if I were to do it.

I don't understand the pkg binary package process. I don't know where it would be documented. But I observe that after a patch release that there is always a gap of pkg service for many packages that take longer to build. If I wait a while then they will appear and can be installed.

I presume that's one of the reasons for using a local pkg build poudriere system.

freebsd is very usable as a desktop; good support and excellent performance

I ran it as my work desktop for a good number of years in a sea of windows

binary packages are also quite viable these days

or ports

I know it depends on what's installed, but how long does it take to update if using all ports?

I would prefer to not build FF/Chrome, even if it's just once for a custom repo/sharing

However, after building chrome I got exposed to ninja/gn and adopted it for our work buildsystems, so :)

it takes a while, but i usually set the BATCH and let it go overnight. it's not like it needs constant attention

also: what's the rush

older machines/laptops

it's otherwise very lightweight (or can be)

having said that, the build infra is very well designed

KMM, I run FreeBSD as my main desktop. I reinstall "pkg upgrade -f" after major upgrades and I am always impressed at how fast EVERYTHING is reinstalled as a binary package. I don't have a time for it but if I were to reinstall everything on my Debian system it would take an hour or so grinding away. But on FreeBSD it all happens in what I recall as just a few minutes.

KMM: there is no need to build ports manually. unless of course, you need something customized.

KMM: that said, just install your pkgs with "pkg install"... similar to linux distros

apt is quite complicated with lots of edge case handling which makes it slow. On my Arch system updates take a couple of minutes.

as does fbsd of course