Re^2: File handles in regular expressions

[Lotus1]

Hi. Did you choose to use Tie::File instead of just reading file1.txt into an array so that very large files could be handled? If so there is a problem with concatenating all the output into a scalar. $matched_lines could end up holding the whole huge file. One possible solution is to replace $matched_lines .= $match.$/; with print $fh $match.$/; Just print incrementally.

[2teez]

Hi Lotus1,
If so there is a problem with concatenating all the output into a scalar. $matched_lines could end up holding the whole huge file. One possible solution is to replace $matched_lines .= $match.$/; with print $fh $match.$/; Just print incrementally.

Not so, am afraid your suggestion will further affect the performance of the script, because the print function would be call as many times as the strings matches, meanwhile with the scalar used no call is placed.
Using a Profiler (NYTProf) made that very clear.
Try it.

[Lotus1]

If you run out of memory the performance won't be so good. I was questioning the logic of using a tied file while holding all the output in memory. The OP didn't state the file sizes or performance requirements. But since you seem to be focused on performance wouldn't it perform better to read the file into an array? For larger files the tied file will only keep part of the file in memory so it will end up rereading the file many times.