> This seems like a problem in search of a solution. The stuff that > needs compressing is already compressed. The stuff that isn't already > compressed, like HTML files and text files, is small enough that > compressing it is going to take longer than sending over the wire raw. > > It's just not convincing.
I disagree. I think the way things are set up now, a significant portion of a freenet user's time is spent reading HTML and other compressable formats in search of media files. zlib compression is very computationally inexpensive, especially when used with human readable file formats. It can compress text files upwards of 80-90% in many cases. This data does not make up a large part of freenet taken purely as data size, but it does make up a significant percentage of the files requested by users. Compression of text and HTML files will certainly not take longer than sending them over the wire raw. Even when sent over a local loop in HTTP 1.1 with compression, the increase in retrieval time is about 6-7%, the reduced i/o time nearly makes up for the time spent doing compression, and that's without any network lag at all. I definitely think this would help improve the
