hello

I am trying to get the xorg to work

What happens?

I get three little screens

Title: xorg does not work exit error 2 | The FreeBSD Forums

solution is solved thank you

Title: c - How can I make Motif Push Button widget to appear as permanently pushed? - Stack Overflow

hello

how do i change the send recv tcp window size?

send specifically

syle you mean like these kernel parameters in Linux ? serverfault.com/questions/775837/ho…um-tcp-receive-window-size-in-linux

Title: How to set the maximum TCP receive window size in Linux? - Server Fault

20:24:19.484251 IP dandesktop.49695 > router.microsoft-ds: Flags [.], seq 2486412285:2486421245, ack 3006939864, win 1024, length 8960 SMB-over-TCP packet:(raw data or conti

nuation?)

this small window size of 1024 only happening in one direction

In Linux there are wmen_default, wmem_max, tcp_wmen that might impact the window size

I assume there might be equivalents in FreeBSD, maybe you can try that

this is freebsd

I know, but sometimes I find easier to find what I want in Linux and then look for the FreeBSD counterparts

tried them all doesn;t work

#net.inet.tcp.sendbuf_inc = 32768

#net.inet.tcp.sendbuf_max = 16777216

#net.inet.tcp.sendspace = 2097152

one of these should work, but it doesn't

When you open a socket, can you specify the window?

I mean in code

yeah I am using -w option with iperf3

tcp_rmem (Receive Buffer) tcp_wmem (Send Buffer)

router:~ # sysctl -a|grep rmem

hw.usermem: 31717912576

router:~ #

you must be thinking linux maybe?

and its sysctl / ioctl just like #freebsd

yeah i was pasting linux foo

router:~ # uname -a

FreeBSD router.sunsaturn.com 14.0-CURRENT FreeBSD 14.0-CURRENT #0 main-n259877-e300f328fec1: Wed Dec 28 23:16:13 CST 2022 droot⊙rsc:/usr/obj/usr/src/amd64.amd64/sys/MYKERNEL amd64

router:~ #

<uskerine> I know, but sometimes I find easier to find what I want in Linux and then look for the FreeBSD counterparts

in response to that

but what about the MTU..

i mean i could try doubling kern.ipc.nmbclusters in /boot/loader.conf and rebooting, idk

mtu is set at 9000 on each end

i get full 10 gigabit one way, and really slow other way, tcpdump showing only difference between the 2 is the window size

this example tcpdump of slow end:

20:24:19.484251 IP dandesktop.49695 > router.microsoft-ds: Flags [.], seq 2486412285:2486421245, ack 3006939864, win 1024, length 8960 SMB-over-TCP packet:(raw data or continuation?)

example of good end:

20:25:00.177144 IP router.microsoft-ds > dandesktop.49695: Flags [.], seq 1533983546:1533992506, ack 176708, win 27247, length 8960 SMB-over-TCP packet:(raw data or continuation?)

so must be window size, i grabbed these dumps from running iperf3 with and without -R flag

so what happens is when i upload a file over samba from my desktop, its 10 gigabit speed, but downloading I get barely 100MB/s on chelsio card, when i disable that and use 10 gigabit intel card i get 300-500 MB/s, improvement but should be full 10 gigabit

what is generating SMB-over-TCP packet?

iperf3 it seems

choose another non-used port for iperf tests

does iperf uses SMB-over-TCP? that sounds weird

yeah sorry that was my actual samba dumps with tcpdump

ok with iperf3 regardless: bad direction:

maybe Samba is setting the window parameter of the TCP packets

because that window is a field of TCP packets, so it can be chosen by the application

23:39:43.819457 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto TCP (6), length 9000)

router.5201 > dandesktop.65214: Flags [.], cksum 0xb673 (correct), seq 212387841:212396801, ack 38, win 4113, length 8960

good end:

23:40:29.840848 IP (tos 0x0, ttl 128, id 37954, offset 0, flags [DF], proto TCP (6), length 5672)

dandesktop.65241 > router.5201: Flags [P.], cksum 0x087d (correct), seq 700082433:700088065, ack 0, win 32768, length 5632

window length is 4113 vs 32768

what i am trying to fix

so you are running iperf from dandesktop to router

yes

that uses 32K window which seems acceptable

then router answers with a smaller window, maybe because it does not ahve to transmit back, the test is just one direction

so it is just acknowledges

have you specified -w at both ends?

mostly good side is full of these:

dandesktop.65241 > router.5201: Flags [.], cksum 0xf22f (correct), seq 700064513:700073473, ack 0, win 32768, length 8960

23:40:29.840844 IP (tos 0x0, ttl 128, id 37953, offset 0, flags [DF], proto TCP (6), length 9000)

bad side is full of ones with window size of 4113

what do you mean by good and bad side?

bad side is downloading to dandesktop: 0.00-1.00 sec 59.5 MBytes 499 Mbits/sec

good side is uploading to router: 8.00-9.00 sec 1.15 GBytes 9.91 Gbits/sec

but are you running iperf where? you might have been working on this for hours, the rest of us know nothing about your setup

i am just running iperf3 on dandesktop to router with:

C:\Users\dan\iperf\iperf-3.1.3-win64> .\iperf3.exe -w 2m -t 10s -i 1s -c 192.168.0.1 -R

and

C:\Users\dan\iperf\iperf-3.1.3-win64> .\iperf3.exe -w 2m -t 10s -i 1s -c 192.168.0.1

adn what do you run in the router?

the -R download side is the issue, that translates to sendbuf on freebsd router

on router i have just: iperf3 -s

why dont you also set the -w in the router?

and what is 2m by the way?

window size i think you can set only on client side

I am not sure that -w is window size

iperf3 documentation says it is socket buffer size

-w, --window #[KMG] set send/receive socket buffer sizes

(indirectly sets TCP window size)

that is not the TCP window size

oh then it does

I would try -w in the router if I were you

i tried won;t take with the -s option for some reason

maybe netcat allows a similar try setting the buffer

fighting to get this window size to match on receiving and sending ends seems to be real problematic tonight lol

man page for netcat only mentions socket buffers, I am not sure if modifying the socket buffer always modifies the TCP window size (which is limited to 65535 I think)

the length of packet looks good, just the window size is way to small on sending side of router it seems

I do not know if that helps, but it seems that iperf is not setting the window size on the router side

I would try netcat

or I would try reversing the test

i can do that

same results

i;ll try from another server as well that runs linux

same results mean that you try to send from router to desktop and see a small window size

?

(using -w in the router side)

gimme a sec linux is a retard needing more than mtu 9000 just to bridge

hehe

alright got the linux mtu happy, yeah bout same results as windows to freebsd as linux to freebsd box

i'll tcpdump here see if just window size again as well

on linux good side: 00:00:22.108440 IP router.targus-getdata1 > router2.53416: Flags [.], ack 133687678, win 4096, options [nop,nop,TS val 944202605 ecr 1071474893], length 0

on linux bad side: 00:01:11.241846 IP router.targus-getdata1 > router2.37240: Flags [.], seq 121612269:121621217, ack 38, win 4113, options [nop,nop,TS val 2338280191 ecr 1071523993], length 8948

hmm

window sizes similar here yet one side doing horribly

and when i reboot between chelsio card and intel card is a difference of chelsio: 50 MB/s intel 350MB/s, both on sending can saturate 10 gigabit

its all about drivers i guess

hardware choices matter/drivers

oww god idk if i can tune these drivers anymore than i have

HCL

i love the wifi dongles that dont support ad-hoc networking

no tethering for you

for INTEL tuning to get that 350MB/s i disable flow control, and i set nrxds to match cores on server, i try same with chelsio with worse results

I can try with defaults but defaults for chelsio is to do 8 queues no matter how many cores or threads

whichever is less, with intel and melanox it will try to match core or thread count

i figure might be easier to find problem with chelsio card doing only 50MB/s than playing with INTEL getting it from 3-500MB/s to 1000MB/s

but the idea here is yeah figure this out, downloading a 4k movie at only 3-500MB/s is rediculously slow for 2023

i mean i use that bandwidth for backups to :)

:-)

For last 5-10 years i been running all my freebsd, windows, linux VMs under centos linux KVM. I really like vm-bhyve so I have moved most VM;s under there now with NVME, drivers can easily pull greater than 1000MB/s so now just getting 10 gigabit issues worked out, and can happily zfs snapshot VMs daily as i;m developping other things is hope since when i tested suspend resume, its not stable,

but i can get it to work

at least under freebsd current tree, i guess i live on the edge running current, but i recompile kernel with production flags , i just like working with latest features

swap the remote hardware out and prove point

pop intel card in and find out..

prove the theory

drivers/hardware combo....

why HCL's matter

Good setup you have going

i like tidalscale's bhyve SSI system

openSSI compaq/hp pretty much killed

need more ram, need more cpu...just add boxes

no single point of failure

high performance computing and high availability

isn;t that cloud stuff, i just never could get into cloud stuff, why not just buy stuff yourself, set it up, if power usage an issue , a few inverters and solar panels should cover the costs and redundancy for home. For datacenter, i prefer baremetal, I don;t think there is anything like having root at the top level and working yourself down to VMs

i;ll just build a new server and ship it to a datacenter, hoping to do a pcie 5 build in a few years lol

because business went cloud

i remember a time when each business had their own network

and the internet was a collection of these networks

when client---server kicked ass

and mainframe/batch processing sucked

we came full circle...

utility computing when everyone on the edge has mainframe general purpose computer in their pockets?

rent seeking!

supercomputer in their pockets

don't forget space internet with elon musk, world domination soon with crypto payments

and supercomputer in their pockets lol

beam the solar energy down to earth with microwaves...beam the data down from DC from space...heat problem solved...co2 issue solved

put the DC in space, hp has

m$ put the DC in the ocean haha

i think it will be fine for next couple decades, but solar, space, wind won;t be answer 3 decades from now with sun coming to swallow earth, underground and using geothermal will be needed

i mean an asteroid will definately take us out like dinosaurs way before sun swallows us, why i beleive in go to mars i guess lol

what would make an interesting movie is what humans will look like on mars 10 decades from now after modifying their DNA so much to adapt to other planets

The Moon is a Harsh Mistress - Robert Heinlein

hack computers with KALI, hack planets and encryption with quantum physics, hack DNA with CRISPR, defintely going to be an interesting world coming up for next generation

already covered

moon needs to be populated first let alone mars

putting your chickens before they are hatched....

oww they need to setup on moon first to launch to mars thats a given, who gonna invest billions landing on mars when can;t do dark side of moon first. Problem is can we do that without governments trying to weaponize the moon

the people born on the moon are practically alien compared to earth born people...gravity and all

they cant even visit earth

anyway scifi paved the way

i think we may have to go straight to mars, i just can;t see a government not being corrupted by ability to send nuke from moon directly to a country on earth

author of hugo/nebula awards though about all this

their bone density

science

low gravity env

i suggest this goes to #freebsd-social but migrating to Mars solves nothing.

haha migrating mars is for fools

it is smaller, it's axis wobbles way more than Earth, means there so dratic climate changes, list goes on and on

before the moon

all this stuff about Mars is just a wet dream and useless money spending

moon was apart of earth once is the theory, yeah don;t want to live there, just bounce off their to mars

yeah cosmic radiation

rna/dna dont like that

well i really think they'll hack CRISPR so much we will look nothing like we do now at one point, so maybe our dna will be ok at some point

one need to grasp, that humans are strictly bound to Earth, evolutionary. nothing more to say

download my being to silicon and upload me into a robot

only other option is to teraform another planet, now that would be alot more work and to many failure points

space travel ...in limbo

i honestly think the solution is travelling faster than the speed of light, only way going to find another earth

cyborg

the speed limit of universe...

you going to break?

🤣

i was talking with NASA last year on their chat channel about it , and someone a month later posted a theory on we could break speed of light if we built a space ship in a vacuum, interesting concept

even quantum entanglement obeys the laws

problem with quantum entanglement and superposition though is that is only works at microscopic levels, we need a human level to

will be great for faster computing and running AI algos, but not sure will help with getting us any faster across universe, although multi universes have recently been proven to exist, so waiting to see what happens with that

is AI really AI though is my question, when i looked into algorithms deep learning uses, its like first year calculus class graph, its a balance of probabilities that if you don;t feed it to much data or over train it it may work

guess would be first year quantum physics class to, idk seems to simple to be the real AI of the future

i find AI would be the next ultimate failure of research ( in physics ), same as useless string theory.

en.wikipedia.org/wiki/Roko's_basilisk you made the AI angry

Title: Roko's basilisk - Wikipedia

dont be messing with a time traveling AGI/AI

yeah who knows, in last year we have learned as facts that big bang is wrong from james webb telescope and multi universes exist, makes you want to smoke a joint and ponder all the 0s and 1s and maybe both at same time in the case of entanglement lol

oh no, multiverse again

another bullshit :)

honestly i think quantum will just be a repeat of history, great they come out with it finally, everyone gets one, takes another 5-10 years for enough developers to do enough useful things with it before becomes mainstream, it will just turn our current tech into mainframes again lol

yeah cause absolute zero in my house cost nothing

governments are only interested in it because of Shor's algorithm which promises to break every encryption right now with just 2k qubits, so all they have to do is keep storing encrypted data and unlock it all in a few years

IBM will have it this year, see what happens

what is compute able,,,turning electricity signal into not just work/spinning a moter but calculating abstracted information...

ibm and samsung duel for #1 portfolio of patents each year now

ibm ran for years as most

well till quantum comes out, asml owns all their asses, hell Biden even trying to ban ASML from selling to china, crazy

the new war is over tech

and i guess thats the real issue with buying even pcie 4 or 5 right now, when we know asml will have chips down to 1nm by end of 2024

dont want russia,china, building drones off latest tech

ok i spoiled myself during pandemic, my desktop is all pcie 4, but thats all lol

rennj they already have it i;ve seen it, it looks like a raspberry pi, same size, but all pcie 4.x , tensor chips, sells for 1k

work perfectly in a drone

with a board like that you could command your drone to facial ID someone and kill him on the spot, thats scary, i think only way to defeat them is with EMP

don't be a menace to south central while drinking juice...

way off topic of #freebsd

sure cause my network stack problem be alot funner on a saturday night :)

Title: Imgur: The magic of the Internet

i have and uppercase problem

Title: Version 7 Unix - Wikipedia

stty / tset -e

i boot up into uppercase for some reason

logout/login fixes terminal go figure

only reason i booting pdp11/v7 cause of #unix question

sucker for irc questions

reset and CTRL-L were my favorites to try first on any OS lol

yeah free compiler would be my quest

heh

shit use to cost money ...people forget

oww you talking about CC from solaris and irix days

1995 choice to make ...$600.00 solaris no compiler for x86/pentium mmx/pro or $50 walnut creek cdrom with freebsd/linux and compilers

honestly i won;t pay for software unless its a game, i put money towards hardware lol

solaris 2.5.1 which i still got

oww god i remember that , i had fiber array controllers on the sparc and veritas backup software, had to work in a datacenter in those days lol

yeah vxvm and vxfs

lol

before sun disksuite/svm

i had irix workstation accessing the SUN;s lol

hp and sun paid them

ibm/hp lvm came out eventually

logical volume management...besides raid levels

and fs

think my first programming language was pascal, unless you count the commodore 64

mos6502 asm is nice

turing complete article other day...

besides x86 mov memory-to-memory being turing complete

mov fusciator

of course programming never complete without doing recursion in assembly and people seem to turn right into sys admins right after that course lol

Title: GitHub - xoreaxeaxeax/movfuscator: The single instruction C compiler

nand2tetris now we need nor2tetris

make some $$ making that book

i;m going to ask the chatAI bot to disassemble this back to GCC lol

nad and nor are UTM

nand2tetris book/project...but no nor2tertris exists

seems cool, should avoid malware detection programs for awhile, i actually heard they were using rust for awhile lol

memory safe languages

foof

should i pseudo-clean install my system

underflow/overflow, Null Dereference

well they just want to cleanup the linux kernel with it and remove all the buffer overflow exploits all over the place in there still, will take a long time

yeah reminds me of old phrack magazine days with alpha one

well asm, c, pascal, joke back in the day about a "gun"

shoot yourself in the foot

building a gun in asm,c, pascal

either way memory safe language are bonus..decades being proven people will make mistake...human error

the null is billion dollar error

according to creator

it has been said the next war will be AI creating its own exploits and hacking servers on its own

i mean you always have the advantage on the offensive, its an infinite war lol

world has changed, in 90s if you ever got known as a blackhat you were finished, couldn;t land a job anywhere, do not pass go, get a job as a bartender. Now they;ll come running to your door and give you free pot i guess lmao

in the late 90 you said html/http and you had job..let alone /etc/services

everything is a file but bsdsockets...ip and port

oww thats right, i said the world php one time and some webhosting company gave me my first job on a solaris sparc programming and admining it lmao

whole point of plan9 was fix things

yeah i guess the first job was the hardest, then i think i controlled the whole countries dialups after that, that was fun

i think my first modem was a 14.4k modem then 28.8k then i was so excited for the 56.k, then cable company came out with cable internet, never looked back lol

i would use them to dialin to univerity systems and run slirp for free internet at home lol

just some old Dec Alphas and Vax systems

everything is a file but bsdsockets do explain

i didn;t do much kernel hacking with freebsd, just did a few programs with kqueue and modified kernel here and there to my needs about it

cat -v is considered harmful perhaps

oww interesting

bsdsocket is not a file..but a ip:port...

Title: A layman's explanation for "Everything is a file" — what differs from Windows? - Unix & Linux Stack Exchange

that would have been useful stacked with netcat for reverse shells back in day lol

open read write close foobar..libc

plan9 tried to fix it

sun went reverse and tried to add streams

everything is file is unix, everything is stream is m$/vms/openvms camp i believe

back in day eveeryone was excited over pthreads, then epoll and kqueue, then everyone said fuckit use libevent or libev to many portability issues lol

unistd.h posix

yeah well smp and smt

thearding locking context switching....

yeah cause context switching was expensive

my sparc boxes could not play mp3's

hehe 70Mhz 64MB-128MB ram, 500MB to 1GB drives

sparc 10,20,5,4,...doesnt matter

sparc20 was only good for 250 nfs connection as nfsserver

its in the documents

dual cpu or quad doesnt matter

i actualyl was going to try to toss solaris again in a bhyve VM other day for fun, i couldn;t beleive i have to use a 32 bit version cause they only let people have 64 bit for sparcs still , i was like fuckit lol

i got 2.5.1, 8, 10, 11.3 vm..why...does it have modern browser

besides the illumos side of things

sparcs were solid though all years i ran, NFS, NIS, veritas , apache etc that system was always rock stable

he who has most developers wins!

and linux is where dev is at

adb-tools! for android

cause phones out number craptops / towers/ 1u /2u /4u

oww get this after i left that job where i got spoiled with expensive sparcs etc, i get to this company and instead of sparcs they have 200 linux 1u servers, every other day i was calling datacenter to clone a failed drive, that was cheapest hardware setup i ever seen

yeah sun had its day..so what!!!

hp had bigger boxes

hp had better engineering..and hpe is still around

did all my firewalling on cisco routers and switches, could never trust any of those linux servers lol

my v2600 had 128cpu/128GB ram in the 90's

sun did not have box like that

not till sun10k sun15k sun25k

starfire which came from sgi/sun splitting the CRAY tech

mandated by U.S.GOV/DoD

my v2600 from hp was on top500 supercomputer list back in day

sun didnt have shit

like m$

winnt on top super computer list...well they tried

supercomputer back then was like i;pm gonna run john the ripper on this encrypted has for next month and see what i get lol

hash

i guess back in those days you literally just ypcat passwd

oww new thing if you like NIS to bad, you have to learn active directory now

redhat now banning it from new systems

so i was sitting here other night ok so toss windows 2022 in a bhyve VM or just run FreeBSD as a samba domain controller, could still use that windows remote tool. Or use redhat new auth system. I;m leaning towards samba with FreeBSD

en.wikipedia.org/wiki/8b/10b_encoding when the signal side and abstract software side mesh

Title: 8b/10b encoding - Wikipedia

when hardware and software work together

Among the areas in which 8b/10b encoding finds application are the following:

haha see the list:

still reading lol on this page now

Title: 64b/66b encoding - Wikipedia

its all claude shannon signal to noise ratio

get the most bang for buck....

ROI/TCO

return on investment and total cost of ownership

"Give me a problem with 1,000,000 lines of C. But don’t expect me to read the C, I couldn't. And don't think I'll have to write 10,000 lines of Forth. Just give me the specs of the problem, and documentation of the interface." --Charles "Chuck" H. Moore

uhdforth does away with spaces!

serious crazy

lol my C professor use to tell me go back to your PC and print out only like the 10 lines of code where you are having a problem and bring it back

oww your program segfaulted, go spend night with GDB and if you can;t solve it come to me in morning lol

i think thats when i started using perl and python tbh

what took me 100 lines of C i could do in 10-20 lines of perl or python

unless you had a reason to close source your code to binary i guess

rennj i might need your expertise to solve this socket issue

im still learning

what side does the heavy cpu lifting when sending/receiving, there is a 2 ghz cpu, i know samba iperf3 generally run single threaded

truss/strace modern day would be dtrace

i guess i could strace the system calls off linux system yeah i;ll try that

dtrace allow you to probe 1000's of times better that trusss/strace

oww wow i thought iperf was advanced

heatmap where kernel is wasting time

its only using select system calls not even epoll

dtrace and zfs pushing tech forward

smf/upstart/systemd...parallel init

zfs is death to fsck...so 100drives being fscked after crash doesnt take forever

omg i think i see the issue but how to figure this out now

read(5, 0x7fd9cb934bd0, 95280) = -1 EAGAIN (Resource temporarily unavailable)

select(6, [4 5], [], NULL, {tv_sec=0, tv_usec=607661}) = 1 (in [5], left {tv_sec=0, tv_usec=607659})

see the EAGAIN

i had boxes take 1hour to fsck raid arrays

that usually means a sysctl that needs to be expanded

resource starvation i;m thinking

yeah for linux i usually use strace, for freebsd i use truss

although they recommend dtrace now yeah

getting closer

ERR#35 'Resource temporarily unavailable'

more like netflix knowing what they can bank on

freebsd redbox on their CDN

dont have to share the code

heh why would you help your competition

build out a CDN and dont have share back...cause the code license says so

w00t mit! etc.

Title: Netflix | Open Connect

Open Connect Briefing Paper: A cooperative approach to content delivery (PDF - English only)

god tracing this down sigh

Title: errno.h in rtems-libbsd/freebsd/sys – RTEMS Project

what sysctl controls adequate buffer send space i haven;t already tweaked sigh

hello

i'm trying to use the desktop installer

it seems to have stuck at "Extracting llvm15-15.0.6_1"

i don't think it should take particularly long, especially i see no disk activity

i'm running it on virtualbox by the way

hmm, the vm froze up completely

does there even enybody exist who does not run freebsd in a virtual environment but on bare metal?

i apparently had a very high memory load in my host machine

wwt, many here do both. Here, laptop running 13.1.

(With some VMs also running 12.3 or 13.1.)

V_PauAmma_V: "running" means without webcam, card reader, dedicated graphics driver and mouse pad?

Running means with a stickered-over webcam (so I can't tell), no card reader slot that I can see so can't tell about that either, but definitely a GPU and trackpad.

And usbconfig says the (integrated USB) webcam is at least recognized.

wondering why desktop-installer is asking for the user to choose the DE even when it is going to reboot the system anyway and thus has to be re-run

apparently it remembers it somehow on the next run, anyway

"somehow" is in /etc/rc.conf would be my guess.

a total guess, but if its for the user ... would it not be possible storing something in ~/

Hi

Did anyone had anything to say from earlier chat?

Really?

daemon, maybe, if the DE only starts after login.

hmm

sixpiece still here?

ahh .. gone

I think I figured out sixpiece's issue with not using the kernel modesetting driver.. you have to 'pkg delete xf86-video-scfb xf86-video-vesa' otherwise.. X will just probe and try to use those first sometimes.. in weird situations even though the kernel modesetting driver will work fine.

sadly they left the channel.. but might be something to add to the FreeBSD Handbook as a caveat?

Patches welcome.

Or if not a patch, a Bugzilla ticket with draft language or an outline. We can take it from there if needed.

despite doing a fresh buildworld/buildkernel/install on my laptop, command-line tools still obstinately refuse to validate TLS certs

browser ✅

curl, git, etc fail

my go-based tool www/gurl doesn't curiously enough

Title: Snippet | IRCCloud

works under dtrace, maaaaybe something somewhere has borked permissions

i am curious now

I have somewhat lame question: when does upgrade to python 3.10 (or 3.11) happen? what is the process?

you can install that package or port at any time, and it should co-exist with 3.9

getting python things to start using it rather than 3.9, that i don't recall

I mean that pkg repository does not have a lot of py310 or py311 packages, most of them are py39. When does switch to the next python version happen in ports?

Question, i have setup poudriere on a extra zpool (not root) data, jail etc are there, but when i update a jail it builds it on zroot under /usr/obj/poudriere (--sysroot) i cant find the setting in poudriere.conf to change that, have i overlooked something or do i have to make a new zfs-filesystem with mountpoint usr/obj/poudriere?

for my curl (and other command-line tool problems), curl loads successfully /usr/local/lib/libnghttp2.so.14

and as root, then loads /lib/libcrypto.so.111 but *doesn't* in userland

so only root can use fetch/git etc with tls

I can read the file easily enough, any ideas what might be missing here?

everything else (as seen by opensnoop) is the same

woop!

just one entry with wrong perms, /etc/ssl/certs//4042bcee.0 was enough

wow

what a timesuck

Question, i have setup poudriere on a extra zpool (not root) data, jail etc are there, but when i update a jail it builds it on zroot under /usr/obj/poudriere (--sysroot) i cant find the setting in poudriere.conf to change that, have i overlooked something or do i have to make a new zfs-filesystem with mountpoint usr/obj/poudriere?

Nixkernal: Check the poudriere.conf settings for ZPOOL and BASEFS

Nixkernal: My setup is similar: bsd.to/TKti

Title: dpaste/TKti (Bash)

afternoon my fellow bsders :)

dch: how did that happen?

does wayland take less resources than xorg?

i would not say too much less but it depends on how you test

it's an eee701

wow that's old

I had an eee701.. but it died.. great little box though.. really portable.. I used to lug it into datacentres and use it as a glorified serial console :D

yeah, finally found a browser that doesn't cause it to melt

wish it could take 12v

disk space is a constant struggle

lynx? :D

elinks, I think, even has support for JavaScript

I think Javascript melts machines

NoScript for the win

i need something graphicall

links does seem to work the best for tet

lynx ui is... not good

w3m-img does show the images (on demand); I think JavaScript is not there

i have used it too

... so is a symlink to a directory a directory or a symlink?

Depends on the usage, access. "ls -L" vs "ls" - if the links is accessed you would get directory; if it is read, you would pointer (loosely speaking) to the directory

s/would pointer/would get pointer/

I've been using uBlock Origin lately, instead of NoScript. I like it more.

I hit bugs with NoScript, also.

i see the error; the script on one host it was a symmlin and the other a directory

rtprio, Perhaps change the script to not care if it is accessing a symbolic link?

ahhh, I need to correct my statement above, tmux's mouse support works in console. I, incorrectly, recalled having an issue the other day selecting windows that where split twice (horz & vert) - but it works fine right now!

parvXirc: i was trying to silence an error about not being able to symlink; problem was directory already existed so the script was woring as it should

zykotick9: yeah, it is kinda neat; i personally find it rather tiresome

rtprio, Ah. Right

rtprio: i was a screen user for years, tmux seems an improvement - the mouse thing is more of a novelty than an actual feature IMO.

How to check if installed world has been updated and aligned with content of /usr/src ?

"uname -a" would reference the commit whichever was used to do "installworld"

s/commit/& hash/