[Toybox] Note to self: singlemake bug.

[enh]

On Sat, Mar 2, 2019 at 8:16 PM Rob Landley <rob at landley.net> wrote:
>
> On 3/2/19 10:09 PM, Rob Landley wrote:
> > make top; make ps; ps -A
> >
> > The -A is a NOP because FLAG(A) is 0 in ps context because ps was switched off
> > when building top, but ps.o didn't get rebuilt building ps. Dependencies wrong.
> >
> > Rob
> >
> > P.S. You wanted more of this stuff going here instead of todo.txt and friends.
> > No idea when I'll get around to fixing it, "make clean; make ps" works fine for
> > the moment...
>
> Also: todo, why the %*#(&%&# is top using 10% of cpu? Ubuntu's top is using 1%,
> I should profile this and try to clean it up.

(toybox top only takes 1.6% for me in its steady state. it's more when
it first starts, but so is debian's regular top. oddly, debian's top
peaks higher at the start, but goes down to 0.7%.)

> (Did the manual stdio output
> buffer setting help at all? Why was that done anyway...)

a little bit, but the real reason for the buffering is to prevent
"tearing" where you get flicker from flushing multiple times in the
same logical update and see half the screen appear before the other
half. it did cure that.

> Also, todo: the unicode fontmetrics stuff was designed on the assumption that
> combining characters go _before_ the characters they combine with, but I allowed
> the elegant sane design of utf8 to cloud my judgement: unicode is insane garbage
> that you have to keep parsing until you find the first character you CAN'T
> display, and only THEN do you RETROACTIVELY know when you've hit the end of a
> line. And if the data comes in slowly, your display device wind up repeatedly
> redrawing the same glyph! That's BONKERS but hey, unicode! What is WRONG with
> these...
>
> Ahem. Todo item re last paragraph: cope. Update design/API to deal with... that.
>
> Rob
> _______________________________________________
> Toybox mailing list
> Toybox at lists.landley.net
> http://lists.landley.net/listinfo.cgi/toybox-landley.net