[matvore]

If we're going to talk about unnecessary extra processes like useless cat, we should merge the head and grep commands into a sed, and possibly just merge everything into perl:

    <access.log sed -n '/mail/p; 500q' | perl -e ...

If perl is processing the file line-by-line then filtering lines by regex and stopping at line X is trivial, and you don't even need sed.

[jkingsman]

I think your point is valid and I don't dispute it, but if I had a nickel for every time I said "just do it all in perl" and regretted it, I'd be... well, perhaps not a rich man, but I'd have lunch covered for a few weeks.

[matvore]

Perl has the advantage of only having one implementation, unlike sed and grep (e.g. BSD or GNU) and /bin/sh (can be one of many POSIX shells), so upgrading this pipeline to 100% perl is safer in some respects. The example in the article is light on details so it's hard to comment very deeply.

I have heard snarky Perl putdowns ad nauseam at work and on HN and may have regretted using it a handful of times but I can say worse or similar for other popular tools, languages...

[b5n]

I usually do it all in awk:

    awk '/mail/ && NR <= 500 {...}' access.log

If you want N matched lines:

    awk '/mail/ && i < 500 {i++; ...}' access.log

[matvore]

I wish I knew awk as well as I know perl, since then I wouldn't need to hear recommendations for CPAN modules and spurious style prescriptions.