Re^2: regex at word boundary

[mikeraz]

The code at that Re^5: regex at word boundary fails on overlapping palindromes, not that I'm aware of any that exist in the wild. Such that adding

nested testest detsen nested
i prefer pi ip referp

yeilds:

line 7:
(0) "nested testest detsen nested"
(7) "testest detsen nested"
(15) "detsen nested"
(22) "nested"

line 8:
(0) "i prefer pi ip referp"
(2) "prefer pi ip referp"
(9) "pi ip referp"
(12) "ip referp"
(15) "referp"

I also tested it on a handy text file of 79,569 lines and it ran much slower than the code I listed above, modified to just test on each line, not each paragraph.

sunorccws04 ~$ time ./mr_pal.pl trf > mr.out

real    1m2.161s
user    1m1.210s
sys     0m0.280s

sunorccws04 ~$ time ./qm_pal.pl trf > qm.out

real    2m53.492s
user    2m49.070s
sys     0m1.690s

trf is the output of a tcpdump session. Other data sets are sure to produce differing comparative speeds.

[QM]

Are you comparing apples to apples? Does the other code find overlapping palindromes?

[mikeraz]

It does catch the overlapping. But I'm not comparing apples to apples because the RE in your code isn't working on my Solaris 5.6.1 code and it's generating many extra matches.