Re: [cgmo-webcgm] miter limit

[Lofton Henderson <lofton@rockynet.com>]

Excellent, thanks Forrest.  I think this should put the question to
rest.  

Rob (in off-list dialog amongst implementors) verified that the
documented PostScript algorithm is the same as the documented GDI
algorithm -- revert to bevel.  Forrest verifies that GDI
implemenation matches GDI documentation.  Forrest also points out a
wrinkle that PS and GDI differ in their *default* values ('miter' join
style set, but no explicit miter limit set), and both differ from the
CGM:1999 default that is essentially "no limit".

So CGM:1999 algorithm ("truncate at the limit") differed from
both ("revert to bevel"), and that is what we corrected in the
WebCGM 2.0 text.  WebCGM 2.0 issue closed.  (Agreed?)

QUESTION:  do people agree that it is a good idea to pursue a
CGM:1999 erratum?  

I think it is an error.  Although this stuff was written more than a
dozen years ago, I can't imagine that the CGM people intended, after
borrowing the concept from PostScript, to define an algorithm that had no
hardware support (in PS, GDI, probably others) and would be expensive to
implement.  To my knowledge, no one has implemented the CGM:1999
documented algorithm.  

-Lofton.

At 04:46 PM 9/13/2006 -0500, Forrest Carpenter wrote:

> All,
>
> After digging into this, I believe the Widows
> GDI and postscript are the same when the values for miter limit are
> actually set in the postscript file and set in the GDI rendering. The
> reason we were not drawing the correct 170 join under the default setting
> was we were not actually setting the miter value in the GDI. It turns out
> the default value for miter limit in GDI is 10 and the default value for
> miter limit in postscript is 14. A value of 10 will result in a join
> bevel at 170 and a value of 14 will result in a miter join at 170. If
> implementers set the value for miter limit in the GDI code the result
> will be correct. The current default for miter limit in CGM is 32767. If
> we can change this to match the postscript value of 14 we should. High
> values of miter limit can result in spikes all over the place in some CGM
> files. I have attached a PNG file we generated from our
> implementation.
>
> Regards,
> Forrest