UNIX for Dummies Questions & Answers

Can't use that option on my solaris:

Code:

grep: illegal option -- P

Oops, sorry. That was the option labelled for GNU grep. Check in solaris for a similar option that says " Interpret PATTERN as a Perl regular expression. "

I think you guys are missing one important point... The longest delay is always introduced by disk reads. When file is being read it is opened as a whole, so any operation that is performed after file is opened won't reduce main delay. And I don't think you will be able to open just part of the file you are interested in. If you are really concerned about those 5 minutes searches, or if you think it might get worse in the future (when number of files to be checked increases), then I would suggest you migrating to some database with text indexes (I think PostgreSQL has that feature).

Can you please explain what do you mean by "opened as a whole"? Not the whole file gets loaded into the primary memory.
If its a larger file, it cannot load everything in one shot and it has to go back, do a disk read and fetch it - page failure
Opening part of the file is not an option here, the OP is interested in searching only part of the record and not part of the file, which means the whole of file ( all the records ) have to be searched but no need to spend time on scanning through the whole record.

I ment reading full lines (records) while file is opened... Let's assume you do have a file with records long enough to be stored in many disk blocks. Lets further assume you can tell system to stop reading disk blocks after certain part of this long record was read. Now how do you tell when next record starts? I remind you that file is a stream of bytes, so next record will start somewhere in next few disk blocks, but where exactly?