Misplaced code

[Lars Hellström]

I'm a bit confused by the following piece of code (from latin.mtx, and also
in latinsc.mtx):

\ifisint{visiblespace}\then
   \setint{boundarychar}{\int{visiblespace}}
\fi

I think I can see at least one point with it---by making visiblespace the
boundarychar one ensures that letters kern to a TeX interword space as they
would have kerned to a space character from a non-TeX-font, because the
visible space character kerns as the space character---but why is it in a
metric file? As I understand it, the visiblespace integer should not be set
until the encoding is determined (as it stores the slot number for the
glyph visiblespace), hence in particular it should never be set when any
metric file is being read. What's more, the location of the boundarychar
must be part of the encoding specification, hence that integer shouldn't be
set by a metric file either.

It appears to me that this particular piece of code really belongs in
t1.etx (and probably in a few other encoding files as well).

Lars Hellström