Failure of Centralization over HTTP
— Please, compress and send the code to me.
— x.tar.gz > here it is.
Today, I started to think that we don’t have adequate tools for source code compression yet. We have great tools for text compression. But what about source code compression? Can we compress source code better than regular text?
Most compression models uses dictionaries, use of common words, relation of words, frequency of letters. But when we think about a software source code, keywords are so few, and the computer languages has a grammar which is simpler than natural languages. Even if the code comments contain a novel we could detect it and use regular compression dictionaries for this part.
At first glance, source code compression sounds awkward and useless effort. But today I can’t access to Github because of the blockage in my country. Even worse, I need to configure all installers using Github sources with proxy… I imagine a real Git-HUB could be decentralized as its origin of the idea and this kind special algorithms could be used in this HUB while transferring chunks of source code between peers. I wish i could have an access to decentralized HUB via my terminal, search what i want as i search in P2P Hub and get what i need from the peers…
Because popular repositories mean more peers, It should be more than a star count in a page. If this kind of Hub is supported by university mirrors, and open-source advocates I believe it could be unbeatable.