3 ms·
Hey, somebody has a patent on pre-loading the compression tables with frequently-encountered strings. So heads up. It was patented in the context of headers fo
by JoeAltmaier 12y ago
Hey, somebody has a patent on pre-loading the compression tables with frequently-encountered strings. So heads up.
It was patented in the context of headers for text messages, which are sent for every message, are generally larger than the message but appear exactly once in each text so never end up in the compression table. Each side of the conversation runs the zip algorithm on the header strings, discards the output up to that point, then compress the text message from that point. On receive they preload the algorithm with the checkpointed data structures from the preload-compression.
This Google technique is probably different enough to be off their radar. But who knows.
I think its the guys who do DirectPC
- makmanalp 12y agoWow, do you have a link to the patent? It'd be surprising to me if there wasn't prior art. Might be worth asking http://patents.stackexchange.com/ http://patents.stackexchange.com/ !
- JoeAltmaier 12y agoCame across it 7 years ago when new at my current project. Did research into efficient compression schemes. Had to change our scheme to avoid the patent. It was Hughes I think.
- JoeAltmaier 12y agoOk I searched patents using 'compression dictionary' and came up with a boatload. Folks have been using dictionaries since 2005 at least, to replace common words e.g. XML tags with tokens. E.g. patent 20050027731: Daniel Revel