[Bugs] [Bug 1149943] duplicate librsync code should likely be linked removed and linked as a library

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1149943

Wade Mealing <wmealing at redhat.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
              Flags|needinfo?(wmealing at redhat.c |
                   |om)                         |



--- Comment #16 from Wade Mealing <wmealing at redhat.com> ---
I do see all very good points made there Neils.


> (Changing the algorithm would possibly introduce continuous healing of data 
> that is already correct, in the case where users have different versions 
> installed - performance impact.)

Right, if there is an implementation problem, I can understand not changing in
mid-release as it would change customers expectations/performance and I'm aware
how that can upset people.

> Adding a function to do sha256 just to replace adler32 does look like a good 
> approach to me.

I'm ok with that, although now the function is a little misnamed.


> The performance difference between weak/strong checksums may not be very 
> relevant anymore?

The new haswell CPU's from intel have increased performance ( see
http://www.intel.com.au/content/dam/www/public/us/en/documents/white-papers/haswell-cryptographic-performance-paper.pdf
) for SHA checksums.

So TLDR for now:

Adler32 is probably fine for now.  Can we / should we get stronger checksums
for the stronger checksumming for a future release ?

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=qkzWQCzEod&a=cc_unsubscribe