Quantcast

Using a faster hashing algorithm for input/outputs.

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Using a faster hashing algorithm for input/outputs.

Luke Daley-2
Hi,

Have we ever experimented with the savings of using a faster hashing algorithm for task input/output files? MD5 is cryptographic, which we don’t need.

Would be interesting to see if there’s a noticeable difference if we used the JVM’s Adler 32 algorithm: java.util.zip.Adler32




Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Using a faster hashing algorithm for input/outputs.

Szczepan Faber
Hey,

Yeah, this would be interesting!

Currently, we do some effort on minimizing the impact of caching by
using a shared cache for the hashes.

Cheers!

On Sat, Aug 16, 2014 at 7:17 AM, Luke Daley <[hidden email]> wrote:

> Hi,
>
> Have we ever experimented with the savings of using a faster hashing
> algorithm for task input/output files? MD5 is cryptographic, which we don't
> need.
>
> Would be interesting to see if there's a noticeable difference if we used
> the JVM's Adler 32 algorithm: java.util.zip.Adler32
>
>
>
>



--
Szczepan Faber
Core dev@gradle; Founder@mockito

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


Loading...