Yet another chapter in the 1.0-to-current upgrade story. :-)

As we saw in last issue's instalment :-), I left "make build" running
and went home.  Today I get here and find that it's croaked:

===> lex
flex -t -p /usr/src/usr.bin/lex/scan.l > scan.c
"/usr/src/usr.bin/lex/scan.l", line 68: bad character: %
"/usr/src/usr.bin/lex/scan.l", line 68: bad character: =
"/usr/src/usr.bin/lex/scan.l", line 68: bad character: "
"/usr/src/usr.bin/lex/scan.l", line 68: bad character: .
"/usr/src/usr.bin/lex/scan.l", line 68: bad character: "
"/usr/src/usr.bin/lex/scan.l", line 69: unknown error processing section 1
"/usr/src/usr.bin/lex/scan.l", line 69: bad character: %
"/usr/src/usr.bin/lex/scan.l", line 70: unknown error processing section 1
*** Error code 1

Stop.
*** Error code 1

Stop.
*** Error code 1

Stop.
*** Error code 1

Stop.

My first reaction is to figure that a previous attempt at "make build"
(under the (broken) snapshot kernel) installed a broken flex.  So I go
back to the SPARC binary snapshot and re-extract flex (and lex, which
flex is a hard link to).

But the binary snapshot's flex produces exactly the same errors!
Perhaps if I could get this flex built, it would be able to rebuild
itself, but at the moment I'm caught between the chicken and the egg,
to mix my metaphors.

For the moment, I'm just removing "lex" from SUBDIRS in
usr.bin/Makefile, but this clearly is not a fix.

Comments, anyone?  Would it work to just get someone with an
up-to-current flex to mail me the resulting scan.c?

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu