[texhax] inconsistency between pdftex and tex

Karl Berry karl at freefriends.org
Sat Oct 18 21:47:36 CEST 2003

    "software systems that are fully compatible with each other";

Yes.  I've seen DEK state (I think in answer to bug reports on
tex-implementors) that obviously the technical part of the trip test
cannot catch extensions.  For instance, if some implementation defined a
new primitive \xyzzy, trip.tex isn't going to know about it, and yet the
implementation should not be called TeX.  Parsing a %& line falls into
the same category.

My dim memory is that another non-technical part of the trip test, that
"implementors are happy with the way it operates at their installation",
is the catch-all for stuff like this.

Anyway, I believe I implemented %& many years ago (or integrated a
patch, or something, memory fails me), and failed to recognize this, so
didn't change the name of the program.  Fortunately, sometime since this
was rectified, and the %&-enabled now announces itself like this:

  This is TeXk, Version 3.141592 (Web2C 7.5.2)
   %&-line parsing enabled.

(There's also an option -parse-first-line to turn it off.)

So Phil is right that the program shouldn't be called TeX -- and isn't.
And David is right in his original bug report, that pdftex and tex
should operate the same way.  I'll mention it on tex-live.

Finally, I should note that mf and mp operate the same way in re %&, but
their names haven't been changed yet.  That will have to be fixed in the
next release.  I'll report that too.


More information about the texhax mailing list