Re: PyDia and Metapost

["Rob McDonald" <robm asdl gatech edu>]

> Wouldn't you know it, the size problem goes away when I strip everything
> down to a minimal example...  Murphy hard at work.
>
> I'll try to figure out where the problem lies.  We all know Knuth's stuff
is
> bug-free (TM).
>
> However, the vertical positioning still seems to be off by about the
height
> of a lower case letter.  Baseline vs. character center.

Well, now the problem seems to have gone away entirely.  Here is what I
found along the way...

Initially, I had a moderately complex Dia drawing.  28 objects total.  cmr10
font used, size 0.30.

Output a MetaPost file, the text is included with something that looks
(roughly) like this....

btex {\usefont{OT1}{cmr}{m}{n} Foo Text} etex scaled 1.56xxx

The number 1.56 is a font scaling number, and manually editing it changes
the font size produced by MetaPost.

So, I created a simplified Dia drawing by deleting 27 of the 28 objects in
the above file.  I wrote it out, and the scale factor above became 0.629883.
The text appeared as the right size after LaTeX & MetaPost.

I thought maybe one of the object types was messing me up.  So, I went back
to the complex file (28 objects), and deleted all the connector lines (14 of
them).  I wrote out the file, the scale factor was now 1.14199.  Closer, but
still wrong.  At this point, I had suspicions of the object _count_ playing
a direct role in calculating the font scaling parameter instead of an object
incompatibility.

So, I decided to re-run the fully complex case so I could get the exact
value of the scaling parameter (1.56xxxx).  When I re-ran it, it now came up
with the correct value of 0.629883.

I'll keep messing around to see if I can make this problem more
reproducible.

             Rob