this terminal seems to have an option for "current loop" serial...

does that mean i can use a RS-422 serial adapter?

RS-422 is a current loop after all

topc what do you men actual xterm? like a physical device?

johnjaye, Years ago an X Terminal was a display and keyboard but instead of having a full computer it was a graphical terminal which would login into another computer using XDMCP protocol (the equiv to a serial protocol for a serial terminal). They were less expensive than a full computer system.

X still fully supports this combination.

yeah but that's a general terminal

i didn't think an xterm was a literal type of terminal specific to X11

if i understand you right

That wiki page and image was not of a "general terminal" like it was a VT100 or anything. It was a graphical X Terminal. Which is specific to X and XDMCP.

If a VT100 uses a serial protocol to connect to a server then an X Terminal uses XDMCP protocol over the network to connect to an X server on another system.

And an XTerm is a program which emulates a VT100 serial terminal. Which is why we describe xterm (and all of the others) as a terminal emulator.

You are right that an X Terminal is a general use terminal because it is just a graphical terminal. But they were never that popular. Rather rare. And today you might see 50 serial terminal museum pieces like hjf's one before seeing an X Terminal. Rarely seen anywhere even when they were new.

rwp: so this worked. i painstakingly remapped all keys to known ansi sequences and changed the termcap

it's not ideal but it's something, lol

the only problem with this approach is that it only works for caps lock off

i've been reading about this and esc [ O D is related to application keypad mode. at least that's what chatgpt suggests

hjf, There is an old adage. "If it is stupid but it works then it is not stupid."

so almost every day I type in 'freebsd-update fetch;freebsd-update install;pkg update;pkg upgrade' and if it completes fine, that's good... if not I interact at whatever step and maybe redo. The problem is, it only autocompletes for 'freebsd-update fetch' then after the semicolon won't autocomplete the rest... is there any shell syntax/setting/replacement I can get autocompletion for the rest

fstty rows 24

gah

it refuses to work in 25 line mode

darwin: that's dependent on your shell's completion features and/or files

it's just root default

ok it seems to have one more bug left. for some reason the words in the last line repea words in the last line repea words in the last line repeat

no idea what could be causing that

darwin: Then it should be handled (or as it seems in your case, not handled) by csh's completion features.

i.imgur.com/G4NE1vK.jpeg

hm that extraneous S at the end of my link appeared when i pasted the link on the desktop machine

when i detached screen on the terminal, it pasted without noise

hjf: Do you want us to look at the image or is this about the extraneous 'S'?

hjf: nice

rwp that is a fine piece of museum! I wish I could have one of those, alghouth the 80x25 is pretty limiting

rwp hjf that is a fine piece of museum! I wish I could have one of those, alghouth the 80x25 is pretty limiting

There is this bug for the r commands: forums.freebsd.org/threads/rcp-issu…commands-package.95211/#post-674511 which if Iunderstood correctly affects to the base system. How do you submit a bug? Is it realistic to think that someone would fix it?

is there any way to get mysql 5.7 on 13.3 or even 14.1 ?

you can try to build it

but it's DEPRECATED: Upstream support ended in October 2019

so there isn't a package anywhere for 5.7.44 ?

if package exists, it will depend on other packages which are outdated/nonexistent

but you can try to build it

it's ok to move forward with the times but there should be a long term pkg repository somewhere

rigth now I have an older server running 5.6 and I can't move directly to 8.0

blame FreeBSD's great stability for running so well it didn't need an update in so long

I agree with you last1, packages shall be stored forever so you are allowed to stay in one version. If it is not broken, why move forward

133 GiB for current snapshot of packages is required

maybe you can find 5.7.44 on some outdated mirrors, but installing it will not probably solve the problem, a lot of dependencies were updated in the meantime

I located mysql server 5.7.38 pkg

installed all current dependencies for 13.4

and just did pkg add and it worked without a hitch

cool

if there's any C++ people who can help with this berthubert/simplomon #16 I'd love to get it into ports but I can't figure out what needs to change

dch: I'm not much of a C++ person, but get confused prior to that point when reading issue 16. Are you able to build that 1875ec4 commit which is mentioned? Does the following commit fail?

|cos|: it fails after this point

hi folks, just to let you know that in Proxmox VE, the GUI interphase the mice will not work

other than that, its all cool, hence do not install DE

I can't think of any reason to do it in the first place, so... :°)

yes I will utilize a hard disk

Have any of you used ZFSnap?

What is all this nonsense?: releng/13.4-n258260-bc3877972eb

you're going to have to be more specific

FreeBSD shodan.trioptimum.com 13.4-RELEASE-p1 FreeBSD 13.4-RELEASE-p1 releng/13.4-n258260-bc3877972eb SHODAN amd64

My 'uname -a' output.

right, so you're trying to understand the composition of the version string?

Pretty sure it didn't used to have thatbit after 'releng'

hmm, good point

my 14.1 build doesn't, I wonder why we changed it

you mean the n258 part?

that's more like the stable/main format

The part I first pasted.

Version strings like '13.4-RELEASE-p1' I'm very used to.

CrtxReavr: it tells you what commit was used to build this, bc3877972eb in this case

What's n258260 then?

CrtxReavr: the first bit is the git branch, the middle bit IIRC monotonically increasing commit #, the last bit the commit built from

in case I'm getting it wrong look in cgit.freebsd.org/src/tree/sys/conf/newvers.sh

some of this is added to help with reproducible builds

some of it was added to appease people who are used to svn and afraid of git

Makes sense. . . things can get wacky with git.

I gotta say though, I do miss the convenience of 'sudo svnup release'

git pull --ff-only upstream/main is hardly more complicated

aah I "misspoke"

--ff-only?

dch, the annoyance is in setting up what upstream/main means for the system you're on.

n258... is count of git commits since the first parent (like a tag or something I think) via `git rev-list --first-parent --count HEAD`

ff-only means "don't rebase, just fast-forward from here to there"

avoids weird merging

CrtxReavr: I see. then you might want `net/gitup` instead man.freebsd.org/gitup

you'd still need to modify the gitup.conf file, but its pretty straightforward

much simpler than dealing with git

This VPS I'm on was deployed as v7 (I think). . . it has 40GB of disk.

It's getting harder and harder to maintian with as FreeBSD has aged over the years, as far as disk space constaints.

dch: should be since the root, and --first-parent means we don't get lost down merged in branches

it's the solution to hashes being useless to tell if your current build contains some EN/SA

16G /usr/obj

That sure doesn't help.

203M /var/db/pkg

198M /var/db/portsnap

Nor does that.

ouch its been so long since I used ufs that I have forgotten the pain of partition constraints

40G of disk back then must have been massive

I wouldn't say massive. . .

I remember needing to get exec signoff to buy 9 3GiB disks

I tempered it for the ask of the VPS. . . ie a place to IRC from, play DNS server, and do a little low-volume web hosting.

s/ask/task

today has gone into figuring out that linux & freebsd behave differently when assigning broadcast addresses over dhcp

I will now go and find a beer to drown my sorrows

I remember first time I used a hard drive, vs. shuffling floppies. . .

a 20MB HD on a Tandy 1000 8088.

those 8" floppies were so cooooool

The HD's seek time was actually slower than floppies.

Still beat shuffling floppies though.

First PC I bought for myself was a PackardBell 16MHz 386sx with 1 MB of RAM and a 42MB hd.

I upgraded it to 2 MB of RAM. . . never ran out of disk space.

Then I upgraded to a Gateway 2000 486dx2/66 with 8MB of RAM and a 340MB HD.

I later added another 8MB of ram and another 540MB HD.

CrtxReavr: With a Turbo-button, presumably?

Yes, though apart from monkeying with it a bit when I first got it, I never used it.

There was a program that let you specify a percentage your CPU to allocate to programs. . . which I used for a handful of games that ran too fast on it.

Actually, I eventually upgraded that computer to 40MB of RAM and added a 2.5GB HD.

curious, do any of you build all the ports with poudriere? just wondering how long that takes. my ports list results in 776 builds, and that's around 18 hours to complete.

I should also say, that PC had a 'VESA Local Bus' which was way supperior to early PCI.

markmcb, arey ou passing -jX to your build?

yeah -j4 i believe

How many physical cores?

8

Do -j12 then.

Unless you really need a big chunk of CPU for other tasks during the build.

it's rust, firefox, and llvm that make up most of the time. other things are quite quick to build

I find 1.5 X cores is a pretty good sweet-spot for build time optimization.

do you build everything?

I used to. . .

Now I mostly use packages 'cept for things I need to customize.

I've been thinking of trying building everything now that "PKG_NO_VERSION_FOR_DEPS=yes" is a thing.

YOu also gotta realize, for the longest time, binaries were built for i386 and if you had anything more than that, you wanted to rebuild to take best advantage of your hardware.

Not as critical these days.

I build my own because I get updates faster than on "latest"

though now that i say that, i should watch it again. because maybe the official pkg build pipeline is less clogged with PKG_NO_VERSION_FOR_DEPS=yes

interestingly there is no such directory called /home hehehe

how i declare some storage partition during installation

By default, that becomes a symlink to /usr/home/ - which is savagely disgusting.

I'm actually quite disgusted that FreeBSD has defaulted to the "big giant /"

For so many years, I used to just hit 'a' for auto partitioning/slicing, then make few tweaks, depending on the available space and task of the box.

hi guys, i'm having a problem with a text terminal. it does this: i.imgur.com/G4NE1vK.jpeg (see last few lines). i'm not sure what could be causing the issue, if a bad serial port, a bad termcap, lack of xon/xoff, incorrect communications parameters (parity etc) ...?

its hard work with installation, i was successful and then i was not

but i guess it worth the efforts

hjf: that could just be an issue with irssi, maybe it's making assumptions about your (custom) termcap entries

do you get the same kind of behavoir when you read a manpage or use an editor ?

s/behavoir/behaviour/ sorry about that

there's also this extra backquote character: "`hjf> no idea what coule be`hjf> no idea [...]", not sure what's going on there

doesn't seem to happen here: i.imgur.com/eZqPwuw.jpeg

nice vintage terminal btw :)

hjf, It might be too much work but I might approach it by writing a diagnostic program which exercised the terminal's escape sequences and tested the behaviors.

I will ask here before I mail the mailing list, I have brought it up before but nobody else reproduced it, if anyone here uses wireguard, does your tunnel break after a network change, aka ifconfig wlan0 down; ifconfig wlan0 up would cause your tunnel to break until service wireguard restart, just me?

hjf, Because right now you have irssi and a screen which is not correct. But don't know why. And it is going to be hard to debug. But if you had a test program which exercised each thing individually then you might be able to figure it out. Move the cursor around the screen. Scroll sections of it. Clear sections of it. That type of thing.

polarian, What network device is wireguard using? If it is wlan0 then I think it expected that taking the device down would cause wireguard to error out.

hjf: maybe script can help. Maybe try something like script -c irssi, then /quit, and try to make sense of the typesecript file

if you get duplicated strings in there, that's probably an irssi thing, otherwise it's likely a termcap issue

rwp: I am confused what you mean by that... wireguard changes the default route so all packets go through it instead of any other interface (unless another explicit route is added)

I am confused by why this is a confusing question. Bringing an interface down almost always creates an error for all other things bound on it. But I have not so far yet set up wireguard myself so I don't know how it is operating.

Hello , I try git pull -C /usr/src like in handbook but is not work , I make # git clone --branch releng/13.2 git.FreeBSD.org/src.git /usr/src

Oh mistake releng/14.1

/usr/src #git pull -C /usr/src not work

# git -C /usr/src pull

Or just cd to /usr/src and run 'git pull'

ek thanks it work

rwp: wireguard isn't bound to it

wireguard is its own interface

the routing table passes packets through the table

s/table/interface/

the default route is via wg0

when it is brought down, the default path is unavailable, packets are dropped

but its weird, if I start a ping, it works, I drop wlan0 so no packets go through, I then bring wlan0 back up and then a few packets make it through wg0 and then nothing, wg0 needs to be destroyed and created again

polarian: some routes get lost in the process

if you recreate them, you will no need to destroy wg0

s/no/no longer

mzar: ah I didn't think about that... but the routes are in relation to wg0 not wlan0

so dropping wlan0 shouldn't cause issues

I will investigate the routing table though, see if it is an issue there

oprs: well there is another layer, it's running inside screen too

rwp: ^^^

but, when i run nano, it does something weird too. the last few chars of the bottom menu repeat in the editor area

so i don't think it's particularly either an irssi or a screen issue

as it happens in "bare" nano

as for it not happening in my previous screenshot, it's because i was using a different termcap for that one

or iirc, when i took that photo, i was actually connected to linux and ssh'd into freebsd

i'm now connected directly to bsd on a serial line

via USB. i've ordered a pcie-serial card though. but it's gonna take some time to arrive

the termcap i'm using was derived from terminfo

Layers and layers! I am enjoying your adventure getting this working. :-)

infocmp -C vp3a+

# Reconstructed via infocmp from file: /usr/local/share/terminfo/v/vp3a+

yeah tried to remove layers but this made it worse, LOL

For serial lately I have been using USB-serial adaptors. A lot more accessible then PCI-e adaptors.

i found an interesting pcie one with a big db25 port, but it's actually an "octopus" cable that splits into 4 DB-9

flashbacks of cisco 2501 as a modem server ***

In addition to serial terminals though a LOT of industrial equipment uses (used?) serial control. Temperature controllers and sensors in ovens. Voltage control of power supplies. A lot of those are serial control and also needed multiple serial ports.

rwp: also if you think this is the weirdest terminal i've set up you're in for a surprise

And also I have in my equipment stash several serial controlled power outlets to turn outlet power on and off. Almost forgot about those.

Back in the day we set up a burn-in oven for life testing parts and all of everything on that system was controlled via serial ports. It worked rock solid. Easy to do and robust.

I look forward to your surprise! :-)

😎

That's a new one on me. I have never seen one of those before.

it's a commodore 128

notice how, while older than the hardware terminal, it's actualy VT102 and it supports colors

I thought we were looking left at the Phillips terminal. I never used any of the Commodore series. I know they were well loved though.

the c128 is pretty cool too. it Ctrl and Alt

it has*

oh that's just a VGA monitor

i don't have a proper monitor for the C128. too expensive nowadays

the C64 is also being a terminal, connected to a BBS. using PETSCII color

So that's a C128 next to a C64. Gotcha.

commodores were actually amazing for their time honestly

I never touched one. I know nothing about them.

this is GEOS, a graphical OS for them

you'd have to spend several thousand on a Macintosh to get a UI like that back in 1985

but you could have that in a $200 C64

or you could have used X too, and spend $10k in a terminal for it, not counting the microcomputer to run it, lol

anyways, back to my original problem

i think it's a thing with the termcap entry

:..sa=%?%p1%p2%|%p3%|%p4%|%p5%|%p7%|%t\E0%'@'%?%p1%t%{17}%|%;%?%p2%t%' '%|%;%?%p3%t%{16}%|%;%?%p4%t%{2}%|%;%?%p5%t%{1}%|%;%c%?%p7%tD%;\E)%e\E(%;:

what does that even mean

chatgpt suggests it's for setting the video modes

bold, underline, standout, dim, blink, reverse

Oh, you're talking about 'Migas.

i'm not

exit

offtopic but I just saw a poor guy making a legit question about a ML algorithm in SO and three guys closed it because "it was not software related".

uskerine: who needs SO anymore? chatgpt has all the answers

It is used much less now than before, but still useful here and there.

well, i have no idea what's wrong with this terminal. i tried other termcaps and the error happens in different parts of the screen. i tried reducing the speed to 9600 and it also happens in a different part of th screeen

the weird thing is that it's repeatable. that repeating quirk happens at the same place every time

i was thinking it could be related to comm settings. this terminal is a bit weird. it's always 8 bits, with parity, and 1 stop bit

i read POSIX doesn't have such modes. One way to workaround it is to set the terminal to "mark parity" (parity always 1), and set the host's serial port to 8N2

What baud rate are you running? Is this using hardware handshaking?

Always use hardware handshaking if available. My DEC VT-100 can't run faster than 1200 over it experiences buffer overruns.

s/over/or/ it buffer overruns. And on mine it is only XON/XOFF available only which makes it inconvenient.

19200 yolo

i tried xon xoff but even after enabling it with stty nothing changes

i'll solder a couple more wires and see if cts/rts makes a difference

If you have hardware handshaking available then definitely use it. If the signals are there then add the wires for it.

Otherwise turn the baud rate down to 300 and test at that speed. I can't know but it really feels like that will be the problem.

i would not imagine that crs/rts would make a difference

rtprio, Without hardware handshaking the server end will write characters to the old terminal faster than the terminal's serial port can process them. This will cause buffer overrun. Some characters will be lost. Escape sequences are always closely packed and most likely to be lost.

On the other side of things typing on the terminal will always be slow even for the fastest typist. And current servers usually have 16550 UARTs or better these days with an internal buffer allowing faster character processing.

You will never see a problem going to the server. But old terminals processing characters from the server? Yes. Most definitely. My DEC VT-100's fastest baud rate without hardware handshaking is 1200 and any faster and the screen becomes garbled.