How I can delete 100 lines anywhere in a file without opening a file and without renaming the file.

I could think of this: You can use this to get a block of lines from a file, deleting the rest of the lines (because of the >).

This works in ksh, but is not portable in sh and bash. This has to do with the way ksh treats the '>'. There is a post where Perderabo has given an explaination for this - but I cant seem to remember where that post is...

Actually I have 8GB file which contains 200,000 records and I want to delete first 50,000 records without opening the file. At the same time I don't want to create another file once I delete the records because of space problem on server.

Is there any way I can delete 50,000 lines from 'abc.dat' and redirect output in the 'abc.dat' only.

ed can do this but it does create a temp file.

Records are potentially very different from lines. You could have your own delimiters and have a record spanning several lines. If each line is a single record, then you could still run the following command under ksh: Though I really dont know if you can do this with such a large file. Even if it works, it will take a really long time to get through the file.

It's not working....it's deleting the whole file..

Are you running this under ksh? It will *only* work under ksh and *not* under bash and sh.