Re: Continue mode in fish protocol - change ibs?

Alexander V. Lukyanov wrote:

>On Tue, Nov 09, 2004 at 02:29:05PM +0100, Piotr Krukowiecki wrote:
>  
>
>>When using e.g. get -c in fish dd first uses a lot of cpu doing nothing
>>(in fact seeking in the file).
>>
>>Could it be possibile to increase input block size for dd command?
>>    
>>
>
>Unfortunately, dd pays attension to read boundaries and it depends on
>operating system what the boundaries are. That's why I had to use ibs=1.
>
>  
>
>>The best if the block size was configurable.
>>    
>>
>
>Probably it would be better to use head and tail commands when they are
>available and support byte counts.
>
>  
>
>>Also there seems to be a bug somewhere in the fish protocol handling.
>>Suddenly the output of the file started to be outputted on stdout/stderr
>>(at least it seemed so). Maybe it was because the remote file was
>>growing when i was downloadin it?
>>    
>>
>
>Yes, it is a problem when the file grows.
>
>  
>
I may sound stupid but, why can't the fish protocol send some tests to 
see if perl or python is available, and then, send commands in the 
selected language (using the -e command line option) ? Since these 
languages have binary wise file handling, seeking files and block 
reading them shouldn't be too difficult...