DEC made a good guess when they produced the VT100. It was the first widely used physical terminal which spoke ANSI X3.64.
Remember: Linux is an ASCII system, so even when you've got a 3270 (real or emulated), Linux thinks it's talking to an ASCII screen. The driver nicely maps X3.64 escape sequences, translating them into roughly equivalent 3270 data streams. Now ... as I recall, UTS donated their 3270 driver to Linux, so we *could* talk direct 3270 data streams. But the applications would need to know about the alternate interface. UTS *did not* donate their slick 'ned' editor, operationally similar to XEDIT and which can drive 3270s directly.
Linux on a PC does the same mapping of X3.64 protocol. In that case, of course, the user end is the text mode raster of your hardware. But the PC itself doesn't know anything about escape sequences or ASCII (or EBCDIC).
"ibm3278" and "ibm3279" don't have meaning in the ASCII/Linux/Unix world.It's not some sort of allergy, nor an aversion; it's just that nobody ever defined terminfo/termcap/curses database files for those entries. But they're not really needed because you get the same effect from (e.g.) vt220 and several others.
On z/VM, you can also 'vmcp def graf' and have DIALable 3270 screens. -- R; <>< On 11/14/22 12:44, Donald Russell wrote:
That’s great.😀 I don’t know why the code was written to act differently when running on vm, but it sounds like using the query partition in all cases simplifies the code and makes it more generally useable. The IBM doc that describes this talked about using TERM=linux and you just mentioned TERM=ansi. Other common types are vt220 and others, so I was curious why the TERM value wouldn’t be something like ibm3278 or ibm3279. I used “linux” and the colors worked as expected, though I don’t use many. I use tput to get the color escape codes to use in the text strings in bash strings and the PSx prompts. As for handling geometry changes due to vm cp disconnect/reconnect… Obviously it would be best if the kernel detected that, but if I have to do something like echo 1 > /dev3270/tty1/query_partition That would be acceptable because I already have a process in Linux that starts a 3270/tty when reconnecting…. Maybe that’s where the code needs to be… when starting the Getty, read partition, that works for me because I also stop the Getty on a CP disconnect event. If you’re open to making changes, I’m open to testing them. (On rhel, I don’t have SUSE or uduntu) I have a couple of other suggestions but this is the most important to me. Cheers, Don On Sun, Nov 13, 2022 at 23:21 Sven Schnelle<[email protected]> wrote:Hi Donald, Donald Russell<[email protected]> writes:This Message Is From an Untrusted Sender You have not previously corresponded with this sender. Thanks Sven, Where do I find the source code? Maybe I can contribute to a solution. 😀It's in the linux kernel source, file drivers/s390/char/raw3270.c, line 580: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/s390/char/raw3270.c?h=v6.1-rc5#n580 As a quick test i changed: } else if (MACHINE_IS_VM) { raw3270_size_device_vm(rp); raw3270_size_device_done(rp); } else raw3270_writesf_readpart(rp); to: } else if (0 && MACHINE_IS_VM) { raw3270_size_device_vm(rp); raw3270_size_device_done(rp); } else raw3270_writesf_readpart(rp); so it always uses WSFQ to get the geometry.When you changed to code to use “LPAR mode” instead of diag 120, did the console work properly with different geometries?It did, but i haven't tested whether dynamic resizing works. And it's diag210, the 'diag 120' was a typo from me.That’s the first problem, when the system first comes up, if the screen isn’t 24X80 the wheels fall off the wagon. But the other problem occurs when the console is disconnected/reconnected and the geometry is different. (A very likely scenario as everybody has their favorite settings. I generally use 60 X 140)Let me try this here. I'm normally running c3270 in tmux, so i'm not disconnecting and reconnecting to my session. But i realize that my workflow is likely different compared to running production systems.I can detect when a reconnect event happens, but there needs to be a way to tell the kernel/driver “check the geometry and adjust accordingly”. I’d like to use 3270 mode, but this is a showstopper for me. Without it working for any geometry allowed by 3270 data stream specification, it’s not something I want to implement. 🙁 Why do I want to to use it? It solves two annoying problems: 1 - it supports “dark” fields so passwords are not visible (ssh to another system etc) 2 - it supports the Linux “progress bars” as created by fsck and others. On the 3215, every update is written to a new line; with the 3270 mode, those lines are just overwritten as intended.It can even show different foreground colors if you set $TERM to ansi. Regards Sven---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email [email protected] with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
-- -- R; <>< ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
OpenPGP_signature
Description: OpenPGP digital signature
