Scanner takes a long time to build.

[Scott A Crosby]

Hello. I was doing a test with flex to see if it could build a scanner for
anti-spam purposes....

I have approximately 70 rules, each a regexp.. When I request that flex
build a scanner (with the default flags)... Well, so far, its run for 10
minutes and 20mb and not outputed a scanner. I can supply the source file
on request. (approximately 15kb)

The rules are along the lines of:

--

remove.{1,15}subject
subject.{1,15}remove
reply.{1,15}remove.{1,15}subject
message\ .{0,10}sen(d|t)\ in\ compliance\ (of|with)
Paragraph\ .a.{0,10}2.{0,10}C.\ of\ S.\ 1618
Bill.{0,10}1618.{0,10}TITLE.{0,10}III
S..{0,10}1618.{0,10}-.{0,10}SECTION.{0,10}301
SECTION.{0,10}301
EU\ (e-?mail\ opt.?out|e.?commerce)\ directive
Directive\ 200.32.CE
opt-in
direct\ e-*mail
mass\ e-*mail
e-*mail\ marketing
This\ a.?d\ is\ produced\ and\ sent\ out\ by
increase.{1,15}\ traffic
this.{0,20}(one|1).time.{0,20}(mail|offer)
social\ security\ (number|record)

--

Weirdly enough, the program seems to spend most of its time outputting the
.c file (at about 100k/sec), not processing it.