[NTLUG:Discuss] find, xargs, grep

[Robert Thompson]

 > Can you give a bit more details on what exactly you did when your
 > computer balked?

I don't remember exactly what I did as it was a few years ago. I do 
remember it was on Solaris when it died (don't remember the version 
though). I just stored it away and remembered for the question "If you 
want to search through a list of files" do the "find -type f -exec grep" 
dance instead of the "grep <long list of files>" thing.

Rob T


Robert Citek wrote:
> 
> On Jul 28, 2005, at 8:01 PM, Robert Thompson wrote:
> 
>> > find /path -type f | xargs grep "foo"
>>
>> I've had that command puke and die because the arg list to grep was  
>> too long.
> 
> 
> I don't doubt that a command puked and died, but are you sure it was  
> because of using xargs?  Are you sure you are not thinking of command  
> expansion?  For example:
> 
>   grep "foo" $(find /path -type f )
> 
> The whole point of using xargs is so that the argument list doesn't  get 
> too long.
> 
> Can you give a bit more details on what exactly you did when your  
> computer balked?  A specific example would be ideal.
> 
>> A better version is:
>>
>> find /path -type f -exec grep "foo" {} \;
> 
> 
> That depends on what you mean by better.  In your example a new grep  
> process will spawn and die for each file found.  100,000 files means  
> 100,000 processes.  Using xargs, a new process is spawned for a group  
> of files.  On a recent test I did I got groups of roughly 500 files  per 
> command.  100,000 files means 200 processes, which is a lot  faster and 
> a lot less resource intensive.
> 
> Regards,
> - Robert
> http://www.cwelug.org/downloads
> Help others get OpenSource software.  Distribute FLOSS
> for Windows, Linux, *BSD, and MacOS X with BitTorrent
> 
> 
> _______________________________________________
> https://ntlug.org/mailman/listinfo/discuss
> 
>