Re: time impact of the delta signature

Denis Corbin <[email protected]>
Newsgroups gmane.comp.sysutils.backup.dar.support
Message-ID <[email protected]>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hi,

interim release 2.6.3.RC2 (available in GIT and on ftp.dm3c.org)
provides an enhanced implementation of the block size determination
used to build delta signatures.

In this package, the man page is up to date to get all details about
this feature, but basically the extended syntax is:

- --delta sig:<function>:<multiplier>[:<divisor>[:<min>[:<max>]]]

which encompasses Andy's syntax. "Function" is either 'fixed',
'linear', 'log2', square2 or square3.

These log and square functions are implemented using bit shift
operations for performance at the cost of accuracy which is not a big
issue in this context, following Graham Cobb's suggestion.

- -vt option shows the block size used of the about to be computed delta
signature.

Any testing feedback is welcome!

Regards,
Denis


On 27/02/2019 09:42, [email protected] wrote:
>> Il 26 febbraio 2019 alle 20.40 Denis Corbin <[email protected]>
>> ha scritto:
>> 
>> Thus since that time, I enriched my thought with the function 
>> parameter not being a number but a real name and keep having on
>> or more parameters as number. something like this:
>> 
>> - --delta sig:<function>:<num>[:...]
>> 
>> where function could be "log", "square_root", "cube_root",
>> "linear", "fixed", while <num> would an integer number
>> 
>> - --delta sig:linear:10000
>> 
>> would correspond to your --delta sig:1:0.0001 in other word block
>> size being 1/10000 of file size
>> 
>> - --delta sig
>> 
>> would equal to
>> 
>> - --delta sig:2048 which would be the same as: - -delta
>> sig:fixed:2048
>> 
>> for backward compatibility
>> 
>> following your notation with - --delta sig:square:10
>> 
>> we would have
>> 
>> block-len-size = (S^2)/10
>> 
>> Doing that way is simpler to understand and simpler to sparse as
>> well in dar. The underlying libdar API would also be more
>> readable that way.
>> 
> 
> Yes, this approach is surely more understandable, please consider
> also the possibility to specify a multiplicative factor for
> obataining, for instance, block-len-size = 10*sqrt(S)
> 
> I.e. if you need integers input parameters, something like --delta
> sig:<function>[:<multiplicativeFactor>[:<divisorFactor>]]
> 
> Andy
> 
> 
-----BEGIN PGP SIGNATURE-----
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=TphR
-----END PGP SIGNATURE-----
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.