On Thursday 24 July 2003 14:14, Tracy R Reed wrote:

> > One aspect of this that I DO support is the compression. I think all
> > files inserted into the network should be compressed using something nice
> > and effective like BZIP. This could belimited to exclude certain
> > pre-compressed
>
> I think all data should be treated the same by freenet. It is probably not
> a good idea to make freenet aware of data types other than mime types
> which is just meta data that gets passed along. Let the users and
> applications (which should want good performance also) make the decision on
> what to compress.

That is what I actually had in mind. All the data is handled the same way. It 
doesn't matter if it is compressed or not. But "Compressed" could be just 
another field in the meta data that tells the end client node whether the 
data is usable directly (i.e. without decompression) or of the data needs to 
be decompressed first (if it has been bzip compressed).

Of course, a file uploaded as bz2 would have the mime type of 
application/bzip2 with no compressed flag, whereas a compressed html document 
would have the mime type of text/html with the compressed flag set.

This would give some level of backward compatibility for nodes that support 
compression, and the nodes that don't would "just break" because the mime 
type would not match the data. Not ideal, but it wouldn't be the first time 
there was a "forced upgrade".

Alternatively, ALL files could be bzip compressed regardless of whether they 
are likely to be compressible (according to mime type) or not. That way all 
files get compressed regardless. It helps from the point of view of 
uniformity and there is no need for the extra flag in the meta data, but it 
seems wasteful to compress already compressed data, especially if the data is 
readily identifiable as uncompressible by mime type in most cases.

Gordan
_______________________________________________
devl mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl

Reply via email to