anonymou5 (no.delete@this.spam.com) on July 12, 2020 8:34 pm wrote:
> Linus Torvalds (torvalds.delete@this.linux-foundation.org) on July 12, 2020 4:25 pm wrote:
> > anonymou5 (no.delete@this.spam.com) on July 12, 2020 3:09 pm wrote:
> > >
> > > thanks!
> >
> > You're perfectly welcome.
> >
> > Honesty in advertizing: I've only done this personally for an editor that I care
> > about (not because it's a good editor, but because after using it for too long my
> > fingers are hardcoded), and I needed it to not completely screw up modern text.
> >
> > I literally took the approach of "one unicode code point (aka character) counts as one fixed column".
> >
> > It's wrong, it's complete garbage, but it happens to be complete
> > garbage that works for the limited case I care about.
> >
> > Actually, it doesn't even work for that case - people with
> > complicated names or who have used broken NFD encodings
> > of their non-complicated names will still show right, but
> > because the editor ends up getting the glyph widths
> > wrong - because it really only counts codepoints - you can't really edit them in the editor.
> >
> > Which makes the editor less than great at what it's supposed to do, of course.
> >
> > But it works just well enough that when I cut-and-paste names into the commit message, it gets the right
> > result. I can't edit the really complex cases, but hey, let's be honest: I can't edit them anyway.
> > Once it's outside of the normal Latin1 character set, my personal limitations are the true limiter,
> > and so the fact that the editor I use also doesn't do a great job with them is irrelevant.
> >
> > And I won't even name that piece-of-sh*t editor, because I'm not proud, and nobody else
> > should ever use it. And of all the "you could do this better" approaches, I literally
> > took the absolute minimum "this works for me" one. I felt no need to do any better.
> >
> > So I know it's garbage, and I know it's not right, but I also know that
> > for what it's needed for, it's (just barely) sufficiently good.
> >
> > Doing Unicode truly well is truly hard.
> >
> > But as mentioned, none of the complexity is in UTF8 itself. UTF8 really is a thing of
> > beauty. UTF8 was designed by people both intelligent and smart (Ken Thompson and Rob
> > Pike). The fact that you can count characters without ever having to decode something
> > is just an example of the kind of intelligent attention to detail that encoding has.
> >
> > Imnsho anybody who uses anything but utf8 is simply doing something fundamentally wrong. They
> > may have good reasons for doing it ("legacy"), but they are still very very wrong.
> >
> > Linus
>
> thanks again!
>
> editor you say? for just yourself? hah! interestingly enough
> that's exactly the case I'm playing with myself! :)
>
> stuff like https://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c has helped me somewhat
>
> but then the actual terminal program and the particular font... well... it's challenging...
Handling variable pitch fonts in payment receipts is a pain in the neck. What we do is using functions (libfreetype) to calculate the bounding box of a particular string using an specific font, then justify accordingly and display/print. It is computationally heavy though, you render the string twice. But a thermal printer is a slow thing, so there is no printing performance penalty.