GNU Parallel Bug Reports feature request: no orphan block

[FrithMartin]

Hello:

first, thanks for the incredible GNU parallel!

If I split data into blocks using --pipe and --recstart, it always puts the final record into an orphan block by itself.

Here is an example:

$ seq 1000 | sed 's/^/>header /' | tr ' ' '\n' | parallel -k --pipe --recstart '>' --block 5K wc

     870     870    5112

     854     854    5124

     274     274    1644

       2       2      13

The first 2 blocks are about 5K, as desired. After that, there is only about 2K left, and I would like it to all go in one block. I attach a simple patch (to GNU parallel 20150522) that seems to fix this.

The same issue was brought up earlier:

http://lists.gnu.org/archive/html/parallel/2012-03/msg00072.html

Have a nice day,

Martin Frith

parallel-patch.txt
Description: parallel-patch.txt