[tex-live] teTeX: no next release

David Kastrup dak at gnu.org
Tue May 23 11:03:44 CEST 2006

Hans Hagen <pragma at wxs.nl> writes:

> David Kastrup wrote:
>> Could you give some pointer or clue to the respective error report and
>> symptoms?  The problem is that I am using \lastnodetype quite a lot
>> and it would be quite helpful to me to know how I can avoid failure
>> with this bug short of sending the customer a dozen different versions
>> until one works with the current version of MikTeX.
> if i remember well it had to do with an unitialized variable
> resulting in a negative node number or so; it dates from before the
> token list (reg number >255) bugs (which was fixed)

Does not sound like it.  I reported a bug once where at the start of a
TeX run, lastnodetype was set to 0 instead of -1.  However, in this
particular case, a test for \lastnodetype>13 evaluated to true, and an
immediately following error message expanded \the\lastnodetype to 13
(the latter being the correct value in the situation).  So the problem
was not that \lastnodetype had the wrong value, but that it behaved
like having several different values at once.

> i must admit that i don't document such bugs here and that i never
> used lastnodetype after that, if only because i don't want to depend
> on users updating the binaries (a two year delay -worse when
> sysadm's are not willing to update- is a reasonable assumption)

I am afraid that there are some alternatives where \lastnodetype can't
very well be avoided.  When deconstructing (and reassembling) a list
in \splitdiscards, for example.  \lastpenalty, \lastkern and \lastskip
can't reliably be used for detecting the presence of the respective
stuff on the list.

David Kastrup, Kriemhildstr. 15, 44793 Bochum

More information about the tex-live mailing list