(20/04/2011 05:08 AM)Assassinator Wrote:  The search argument is PB's anyway, not mine.  I do not really see how it is more difficult at all to copy a whole file name and paste it into a search engine, than it is to copy the hash part of a file name, and paste that into a search engine.

I find files sometimes get renamed, eg spaces to underscores or vice versa etc, which means a direct copy doesn't always work.  Also, some search engines suck and try to find keyword matches, so if you only want a particular file, and you search the name, it will return results matching any word in the name.  Won't happen with a hash.

(20/04/2011 05:08 AM)Assassinator Wrote:  But that will be a waste of effort.  Because only the computer is ever going to read the segment ID anyway, no point in making it sensible.

There is as you may need to put in segment IDs, but not much.

(20/04/2011 05:08 AM)Assassinator Wrote:  It's not all that important, even if the transfer protocol doesn't have error detection.  I bet 90% of the people don't even know why it's there at all.

From my understanding, IRC runs over TCP, which does have error detection.

(20/04/2011 05:08 AM)Assassinator Wrote:  ]
But yeah, wee're really going off topic.  The main reason is file integrity checking, I think wee both agree with that.

Word.