Strictly speaking, standout in the #termcap/#terminfo model is not necessarily the same as reverse video.
The Z shell's prompt sequences can only do standout, not reverse video.
Here's a #zsh 5.9.2 patch that adds %O/%o to prompts. (It turns out that the better letters were secretly taken.)
Yes, I have been experimenting with changing standout mode to be boldface+italic. (-:
Should anyone want that sort of thing, I've polished the font conversion utilities in #Unscii and given them manual pages.
They include a #vtfontcvt for non-FreeBSD platforms. (It doesn't do BDF input, but Unscii also has a separate #bdf2hex conversion utility.)
A couple of YAFF utilities, too.
As a consequence, I am building fonts for #nosh toolkit #VirtualTerminals directly on #Debian and #NetBSD, rather than building them on FreeBSD and copying them over.
My proposed changes to the #NetBSDGuide are up. They should, I hope, cater a bit more to the users coming from a world where they know EFI and GUID partitioning, without losing the users who are bootstrapping #NetBSD on an SPARCStation or a PowerPC Mac. (-:
It's a good thing that I took a break and then looked it over again. I made several improvements.
But to prevent perfect from being the enemy of the good, and so that this gets out to all of the potential #NetBSD users reading the #NetBSDGuide sooner rather than later, I'm cutting myself off. (-:
It's in the same place as before. I based it off your latest, so the diff should be fairly clean. However, the Guide seems to be missing some man page macros.
It turns out that #Unscii already had quite a few of the new Unicode 17 characters from home computers. It was just giving them codepoints in the Private Use Area.
So I have spent a lot of today reassigning them to their Unicode 17 code points (in my fork of Unscii).
Since this includes proper octant blocks from the #Kaypro, I've changed the Braille characters back into Braille. Kuhn's sampler now has visibly #Braille spaces, rather than SPCs.
Some more #backspace fun you might like:
#TianoCore makes BS a *destructive* backspace, and that stops various EFI BSD bootstrap loader programs from displaying their spinners properly.
* https://tty0.social/@JdeBP/114658278210981731
Backspace wraparound once upon a time could crash Windows NT.
I've just finished the vtfontcvt(1) workalike for my version of Unscii, and built my first vt(4) fonts directly on non-FreeBSD systems.
I installed them in the various kvt realizers across my machines, and they all seemed to like the new font files.
GNU Unifont has definitely grown since Unscii started a decade ago. It does not even fit in 2 vt font files any more.
https://github.com/jdebp/unscii/tree/2.1.1f
There's a noddy hex2yaff in there as well.
Clearly no-one with any traction at #Debian installs packages in a full-screen kernel virtual terminal in a UTF-8 locale. I can tell because the apt-get 'fancy' progress bar is broken.
It uses several Unicode block characters to draw partial bars. These characters do not exist in the Unicode fonts in the console-setup-linux fonts pack, so the end of the bar comes out as the blob replacement glyph.
I don't use KVTs either, which is why I hadn't noticed until now.
On top of that, to retain the old #login behaviour, Debian #LinuxPAM invokes pam_motd(9) twice, and then pam_mail(8).
This results in three separate calls to the pam_conv callback, which is a problem if one is presenting the #PAM UI as dialogue boxes. One gets three successive dialogue boxes, after the actual #login dialogue box.
In part this is a design failure of pushing two different sorts of things (dialogues and post-login greeting messahes) through one pam_conv function.
#LinuxPAM warts for you:
Consder pam_motd(8). *Unless* the noupdate option is given, it rewrites the motd file by default. Which means that it has to run as the superuser. It rewrites the motd file by running a bunch of shell scripts. As the superuser.
In BSD login(1), the motd is read and printed *after* privileges have been dropped. And rewriting of /etc/motd is under the control of a separate single-shot motd service that runs at bootstrap or on demand.
Since #redo version 1.6 and #nosh 1.42 are out, I've been working on the next versions.
Inspired by a recent conversation about the TUI of #TurboBASIC, I've put (optional) colour support into the TUI login system, and also the ability to use WGL4 as a halfway point between only typewriter characters and using the Unicode BMP proper.
The latter is not an issue for user-space virtual terminals, but the #Debian fonts for KVTs roughly align with #WGL4.
I've also been expanding the redo Guide.
I released #nosh 1.42 yesterday.
There's quite a lot in the list of changes (q.v.), but it's not as drastic as 1.41 was.
There are several bugfixes for things like the case of nvt-client when its standard input is a pipe. In terms of new stuff, there's the gb-shaw keyboard layout, a service to copy the coreboot_console log, more Z shell completions, more programs understanding CLICOLOR/CLICOLOR_FORCE, actual manual pages for those two, and linenumber and sqz after a 43 year wait.
Our @bketelsen has a 'Show HN' on #HackerNews.
#djbwares version 13 is out.
http://jdebp.info/Softwares/djbwares/
There are a few functional changes. Client-supplied data are now vis(3) encoded in logs rather than using the old Bernstein question mark substitution. Non-HTTP protocols are stricter to reject HTTP clients more often. And sntpclock and taiclock can take domain names.
Plus some bug fixes.
#redo version 1.6 is out.
http://jdebp.info/Softwares/redo/
There is not much change to the base tool, apart from redo-delete as mentioned a little while ago.
I've added Z shell completions and the start of an off-line Guide, changed cubehash; and added two helper utilities that I use for building packages.
redo at minimum 1.6 is going to be a build pre-requisite for the forthcoming versions of #nosh and #djbwares, which is why it has been released first.
And today is the day that I got rid of a tool from #redo, replacing it with pkg-gencontrol, before anyone else even knew that it existed.
You ain't seen me, right? (-:
I've given in and added redo-ifdelete to #redo, and it will be in version 1.6 when it comes out. There was one corner case in another package thing that was crying out for it.
Of course, once it was in, I found a number of other places where it was useful. Not the least of which was the build of redo itself.
Don't tell @cazabon, but I made a WWW page about pulling ourselves out of the 1970s, kicking and screaming, into the 1980s. (-: