Web lists-archives.com

Re: Bug#856033: ITP: brailleimg -- produce text images and graphs abusing Braille glyphs

On Mon, Feb 27, 2017 at 03:18:04PM +0000, Jeremy Stanley wrote:
> It's not packaged for Debian yet nor do I see any RFP/ITP, but I've
> been happily using https://github.com/tehmaze/diagram for a few
> years (installable from PyPI via pip so probably easy enough to
> package). Its default mode uses 8-dot Braille patterns for axis
> graphs.

Alas, that tool works on terminals only.  It uses ncurses and its output
can't be trivially converted to text, at least programmatically (you can run
it in a terminal then copy&paste the result).  But even if you do that, you
lose parts that are implemented as attributes (like, a horizontal line is
done via underlining).

And even if you use a graph mode that doesn't need attributes, it does mix
character ranges -- especially bad is using U+20 (ie, real ASCII space) as
on a typical non-terminal setup it has only a width of ~2/3 of U+28XX
glyphs, making everything totally misaligned.

Even on actual terminals, it works only if lowered dots are rendered as
blank.  This goes contrary to what Unicode requires.  Most fonts do so as on
a low-DPI display the text tends to end up in an unreadable jumble of pixels
(see FreeMono before stretch), but with some fancy logic it _is_ possible to
do so in a way closer to what the standard requires.  Please see what Fedora
implements in their default terminals -- when using U+2800 like braillegraph
does you get a white graph in a dark gray rectangle, with tehmaze's diagram
the graph is surrounded by ugly blocks.

Likewise the embedded labels: they're too likely to cause misalignment or
even word wrapping.  A mail or a commit message is going to be seen by too
many diverse clients you can't control.

⢀⣴⠾⠻⢶⣦⠀ Meow!
⢿⡄⠘⠷⠚⠋⠀ I was born a dumb, ugly and work-loving kid, then I got swapped on
⠈⠳⣄⠀⠀⠀⠀ the maternity ward.