Re: [PATCH] dd - support for reading of full blocks

[Jim Meyering]

Pádraig Brady <address@hidden> wrote:
> Pádraig Brady wrote:
>> Kamil Dudka wrote:
>>> as solution to rhbz #431997 and #449263 I propose patch for dd - support for
>>> reading of full blocks. This support is activated with dd parameter
>>> conv=fullblk. This patch has no effect if parameter conv=fullblk is omitted.
>>
>> This feature makes a lot of sense.
>> I'm not sure I like the conv=fullblk syntax though.
>> Here are alternatives in my order of preference:
>>
>> iflag=block
>> conv=fill
>> conv=fillblock
>> conv=fullblock
>
> or conv=wait, though I do have a strong preference
> for iflag=block as it's the most accurate.

Kamil, thank you for the patch.

As for the new name, iflag=block would be my preference if it weren't
for the existence of conv=block/unblock and iflag=nonblock.  Too much
room for misunderstanding.

I'd be happy with either of the following, with a slight preference for
the first (it's shorter and doesn't need the "read" prefix to indicate
that it affects only how input is read):

    iflag=fullblock
    conv=readfullblock

Kamil, with new features, it is customary to write tests
to exercise the new functionality.  One way would be
to do something like this, but appending the new option

    $ (echo a; sleep .1; echo b) \
      | dd conv=block cbs=5 status=noxfer > out 2> err

and then comparing "out" and "err" against expected outputs.