[Standards-JIG] JEP-0138 (Stream Compression): A case for LZO / even better algorithms

Peter Saint-Andre stpeter at jabber.org
Thu Nov 3 21:07:28 UTC 2005


Andreas van Cranenburgh wrote:
> On Wed, Nov 02, 2005 at 01:16:53PM -0700, Peter Saint-Andre wrote:
>> Hi Andreas,
>>
>> Thanks for the information. I don't think JEP-0138 will recommend one 
>> algorithm over another except for making zlib mandatory to implement 
>> (mostly because it is in common use). Some quick research on LZO does 
>> indicate that it is very fast, however, it seems to be a library that 
>> uses multiple algorithms rather than a well-documented algorithm itself. 
>> For instance, see the LZO documentation and FAQ:
>>
>> http://www.oberhumer.com/opensource/lzo/lzodoc.php
>> http://www.oberhumer.com/opensource/lzo/lzofaq.php
>>
>> I think we'd prefer to reference a somewhat standardized algorithm (as 
>> we do for zlib and lzw) rather than just a library.
> 
> If that is indeed true then I'll just invite people to experiment with
> LZO in Jabber [1].
> 
> Thanks for following up, by the way.
> 
>> Am I missing something here?
> 
> Could be, but then again: I certainly am :)
> 
> What I assume is that there's a clear seperation from the algorithm (LZO) and
> implementation (lzop). Eg. note the use of case. But this might only
> be superficial, and perhaps when you look better it's not such a neat
> algorithm, but simply a collection of things. 

Yes, that was unclear to me in the lzop documentation -- it seemed to 
imply that lzop (the library) uses a wide range of algorithms, which 
result in fact compression / decompression.

Even if LZO/lzop is just a library (not an algorithm), that doesn't we 
couldn't or wouldn't mentioned it in the JEP, I just think that case is 
not as clear-cut as zlib and lzw (which seem to be well-documented 
algorithms).

> I suppose it's rather
> adaptive, to the data being fed to it. If Jabber prefers to have
> scientifically peer-reviewed algorithms etc., then I understand that.

I'm not sure what we prefer in this case, which is why we're having this 
discussion. :-)

> How does it work with such standards, do I violate JEP-0138 if I use LZO
> right now in a theoretical python script? Either way, I'll try [1] just
> for the fun of it.

Gosh, no! It just wouldn't be an officially registered library / 
algorithm. The JEP doesn't say that a library / algorithm needs to be 
registered or even mentioned in the JEP. Experimentation would be welcome.

/psa

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3641 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://mail.jabber.org/pipermail/standards/attachments/20051103/25fbf73e/attachment.bin>


More information about the Standards mailing list