On Wed, 21 Feb 2007, John Levon wrote:

[ ...]
> You've lost me. It's always the case that the behaviour on the right side of a
> pipe is dependent on what happens on the left side of a pipe. I can't imagine
> any concept of a pipe where that isn't true.

If the command on the right side of the pipe is supposed to operate on the 
last word of the input string, I just don't expect it to work on the first 
simply because the sequence of words was added to in some way.

If at all, the command on the _right_ side of the pipe should select how 
to interpret the input. If I do:

cat file | awk '{ print $NF }'

I want the last word of the input. And it shouldn't matter at all what's 
_before_ the '|'. If I had wanted the first field I'd have written $0.

I do consider it a bug that I can't take an arbitrary string ending with a 
hex number separated by whitespace from whatever's before, pipe that into 
mdb's ::print and get an output. And changing the behaviour in the way it 
was done doesn't really make it more consistent.

Yes, there's a lot of personal preference in here. I agree my preferences 
may not be everyone's. Note I haven't requested "change it back". There's 
a certain ability to adapt even in folks as boneheaded as myself :)

>
> The fact that the -a option to ::print was previously useless in a pipe 
> context
> is edging pretty close to a bug as far as I can see.

The fact that it didn't cause a difference in pipe behaviour was a 
feature.

FrankH.

Reply via email to