help with qx?

Iain Barnett iainspeed at
Wed Feb 6 13:42:53 GMT 2008

On 6 Feb 2008, at 7:11 am, Uri Guttman wrote:

>>>>>> "RJR" == Randy J Ray <rjray at> writes:
>>> and it will be unlikely that his command will generate enough  
>>> output to
>>> fill a stdio buffer in the qx or the open pipe. so looping over  
>>> the qx
>>> lines in a list context or reading line by line from the open  
>>> pipe will
>>> be about the same speed and storage.
>   RJR> He did specifically say that he expected the command to  
> generate
>   RJR> a considerable amount of output. I suppose that depends on
>   RJR> whether his concept of "considerable" matches ours. Pulling  
> large
>   RJR> diffs or log history on a sizable project could produce fairly
>   RJR> large output.
> if that is the case then open | will be more efficient. but with  
> today's
> ram and buffer sizes, even 1Mb is small. we haven't been told the  
> output
> size but if it is < 1MB (or even more given the ram available) then
> buffering will not matter much. so again, simpler coding should win  
> out
> and qx is simpler than open |.
> uri
> -- 
> Uri Guttman  ------  uri at  --------  http:// 
> --
> -----  Perl Architecture, Development, Training, Support, Code  
> Review  ------
> -----------  Search or Offer Perl Jobs  -----   
> ---------
> ---------  Gourmet Hot Cocoa Mix  ----  
> ---------

I'm quite embarrassed to say it ended up being approx 250k... (gets  

In my defence, the buffer on the Windows command terminal was so  
small that I couldn't tell how much there was - it looked like a lot  
more than that! :)   Still, could be a lot more output depending on  
what is checked out, but you're probably both right about the power  
of the machines etc.

I ended up using a pipe because it still printed to STDOUT with a  
filehandle. Once I got it working I just left it as is.


More information about the mailing list