Hello Bruce,
Hello Hans,

Thank you very much for your help and input. I tested the proposed patch
with the minimal example, and the glitch indeed disappeared.

I have to admit, I wasn’t aware that metafun could run in different
instances, let alone the existence of both a double instance and a scaled
instance. While your patch is quite clear, the inner workings of metapost
are a mystery to me. I greatly appreciate the quick fix and the explanation.

Best regards,
Bernd


---

This version improves readability, fixes minor typos, and maintains a
polite, professional tone.


On Wed, Oct 16, 2024, 22:34 Hans Hagen via ntg-context <ntg-context@ntg.nl>
wrote:

> Hi Bruce,
>
> > If you change the following MWE line
> >
> >      \connection [bt] {compiler}
> >
> > to
> >
> >      \connection [br] {compiler}
> >
> > then the glitch goes away, so I suggest that means it is a bug in the
> flow chart module.
>
> Actually not (but you're close) ... so for a moment put your calculator
> (hardware) hat on (although they use decimal clipping and rounding i guess)
>
> You noticed that coming from one directions gives a different result
> than coming from the other:
>
> metapost        > message : 1.4999979972839355 1.8181943026455964
> metapost        > message : 1.5 2.1818056973544033
>
> Now, why does this fail? We use a double instance here and as we
> calculate intersection points we are too precise compared to scaled
> where we get twice 1.5, so the points collapse in scaled mode and not in
> double. Maybe I need to make some helper for clipping a given precision.
>
> The mp graphic that does the framed loads a metafun double instance so a
> later instance with the same name in the chart (lua) module takes that
> one; when it is the first one it will take the scaled instance (mkiv is
> scaled by default and mkxl defaults to double).
>
> We can run these charts in their own instance (i'll do that).
>
> Here is the patch for mp-char.mpiv that fixes the issue:
>
> def flow_collapse_points = % this can become a core macro
>      begingroup ;
>          % remove redundant points
>          save n ; numeric n ;
>          n := 1 ;
>          for i=2 upto flow_xypoint :
>            % if not (flow_xypoints[i] = flow_xypoints[n]) :
>              if abs(flow_xypoints[i] - flow_xypoints[n]) > eps :
>                  n := n + 1 ;
>                  flow_xypoints[n] := flow_xypoints[i]
>              fi ;
>          endfor ;
>          flow_xypoint := n ;
>          % make straight lines
>        % if flow_xypoints[2] = flow_xypoints[flow_xypoint-1] :
>          if abs(flow_xypoints[2] - flow_xypoints[flow_xypoint-1]) < eps :
>              flow_xypoints[3] := flow_xypoints[flow_xypoint] ;
>              flow_xypoint := 3 ;
>          fi ;
>      endgroup ;
> enddef ;
>
> But best check this! So the bt / tb was a clue indeed.
>
> Hans
>
> -----------------------------------------------------------------
>                                            Hans Hagen | PRAGMA ADE
>                Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
>         tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl
> -----------------------------------------------------------------
>
> ___________________________________________________________________________________
> If your question is of interest to others as well, please add an entry to
> the Wiki!
>
> maillist : ntg-context@ntg.nl /
> https://mailman.ntg.nl/mailman3/lists/ntg-context.ntg.nl
> webpage  : https://www.pragma-ade.nl / https://context.aanhet.net (mirror)
> archive  : https://github.com/contextgarden/context
> wiki     : https://wiki.contextgarden.net
>
> ___________________________________________________________________________________
>