Re: compact - a history-file compaction command

[byron@netapp.com (Byron Rakitzis)]

>Here's my version.  I'm sorry I didn't send it out much earlier and save
>everyone the effort of writing those foul script versions :).  Please
>comment.

Ok, I will.

I think writing a C program for this utility is wrong for the same reasons
that people who aren't fluent with Unix approach such a problem by skipping
to C first, rather than trying the tools

	cat -> tr -> sed -> awk ( or -> perl ) -> C

in roughly that order.

In this particular case, I would much rather implement this history
compactor as

	tac | onlyfirst | tac

since

1) a tac binary is useful in other circumstances, whereas by your
admission a history compactor is pretty much that --- a history
compactor. So why clutter up my bin directory with yet another
binary?

2) it is not much slower than the C program you wrote. And in any
case, I have a hard time believing that the performance of the solution
is a big deal, esp. if you compact the file out of cron. And it is not
so much faster that I would want to use your program, e.g., every
time I logged in.

3) a shell solution is generally easier to develop & maintain, as
discussed ad nauseam on the various comp.unix.* lists. My awk & rc
scripts were written in a matter of minutes.

With those three points aside, I'd like to thank you for the code. It's
hard to argue with an existing implementation that you just need to
compile up, providing you don't care about point #1. (However, I still
think you wasted your time writing it.)

Byron.