Originally posted by zinder
im no DB knaack, so take what i say with a grain of salt. But while your second variant saves DB operations, it is very costly on memory and processing if you want a halfway decent user interface with things like jump-to-entry, search, sort-by-title/date/etc.
Jump-to-enry and search \"very costly on memory and processing\"? I don\'t buy that. If your computer can run Planeshift, it should be able to do a simple search through several hundred kilobytes and return results.
Also, diamondcite, I like your idea. A simple \"diff\" like command that would save bandwidth. I think we could take most of the source code directly off of the GNU diff command.
What if at the beginning of the journal database entry, it would have entries for each of the pages, about whether they are empty or not. The server would only send over non-empty pages, and the client would know that the rest of them are empty. And then, during upload, the client would check if a page was empty, and would upload only ones that are not. Empty pages are pages that contain only spaces and new line characters.
And I just checked out sizes of pages in vi... I think 65 columns by 35 lines, after word wrap, would be great. That is really a very decent amount of page size for everything...