reopen 324308
thanks
#include <hallo.h>
* Michael Stone [Sun, Aug 21 2005, 11:46:48AM]:
> On Sun, Aug 21, 2005 at 04:15:48PM +0200, Eduard Bloch wrote:
> >See output below. It does not make much sense.
>
> Nope, it makes perfect sense. You're asking for a fixed number of
> ridiculously oversized records. It's extremely unlikely that you'll
If that is ridiculous then I wonder why the upstream puts prefixes like
M into the manpage. Oh, and even uses examples with stdin in the info
page, so what...
> actually get 200 1M reads over a pipe; instead, you'll get a number of
> partial reads. The output size is determined by how many partial reads
> you get, and how big they are. This is just what happens when you read a
> specific number of fixed block records from a source that doesn't really
> generate fixed size blocks. If you're just trying to get some specific
> number of bytes you're really better off using head -c. If you really
> want to use dd to count a fixed number of records over a pipe, set the
> ibs to 1 (note that this will be really slow).
Sorry, that all sounds like a bad excuse. I think I can understand the
programmers decission to use such simplistic buffer handling but that is
NOT the documented behaviour. And definitely not what a user would
expect.
Bug apparently to kill the messenger. Do you really want to see this
forwarded to TC?
Eduard.
--
<Alfie> Getty: Solche Aussagen tätigt man hier nicht - sonst kommt
wieder der weasel daher und zitiert einen auf der #-Seite!
