I wish I had known about character encoding before I wrote Tabulas. I need to make a clean shift to UTF-8 (Tabulas is currently using iso-8859-1), but I'm not quite sure how to do this without basically taking the whole site down and converting all the text over. I may do this when I manage to write some sort of compression scheme on the text because the uncompressed text in mySQL is taking up more than a gig of hard drive text. Or should users be allowed to set their own charset? Are there any benefits/pitfalls of this (besides the obvious problems when someone wants to switch charsets and their previous data gets screwed up).

Help.

. . .

It's weird that while I'm developing something, I'll have a cool idea for a new feature (this happens every day), but then a day later, I'll think "that's really stupid." Of course, being incredibly scatterbrained also means that I tend to forget new feature ideas ... which I guess also means I'll ultimately save myself from calling myself a moron...

One of the problems, as I see it, is the lack of linking between entries. If I want to help foster communication, I need to start really examining the entries for hyperlinking information. One of the little ideas I'd like to steal from LJ and wikis is the ability to link to someone else's Tabulas using a command like the in-line image command that Tabulas uses. So maybe you'd type [User:roy] and it'd automatically link to my Tabulas, but the link wouldn't go bad (as it would if you hard-linked it) if my site URL changed. Since I'd like to (in the future) be able to allow users to use subdomains and domains, the site URL may change.. wouldn't it be nice to have your links follow you? Or is that not even the right way to view the web... ho hum.

. . .

In any case, it would be fantastically cool if I could write something that would analyze public entries for links; if the site finds a link that references another Tabulas' entry, Tabulas would store this information. The idea is that: (1) it will let you know who is linking to your content in Tabulas and (2) you can build threaded conversations based on debates.

Commenting, it and of itself, is simply not enough to foster debates of much worth right now. The box is too small, and there are just not enough options. However, if one could write an entry with a link to an existing Tabulas entry that would automatically show the reference on the other end ... that would be incredibly beneficial.

This already exists in the form of trackback/pingback, but what I'm thinking of doing for Tabulas is slightly different. It is trackback/pingback, but it also automatically threads the idea as it spreads around, kind of like a message board. As more people join a specific topic of conversation, the thread grows larger... this could be quite useful in specific cases like Ed's recent posts as I write posts that agree with him while Kris writes posts that disagre; meanwhile there would be one page (somewhere else) that would list the threaded conversation.

It would also be incredibly useful in these "favorite 5 books" memes that are spreading around. How useful would it be to see on ONE page everyone's response to this meme ... in a threaded format? Not only could you trace who started a specific meme, but you could also see random people's responses.

Sorry if this is a bit unclear, but it's just random braindumping.

Not like it'll make it into Tabulas, but it's fun to theorize:)

I'm almost done with rewriting all the old functionality for 2.1. I'll soon start writing the new features (which isn't as tough as a task as it sounds, since my new libraries let me pump out new functionality with amazing efficiency) ...

Posted by roy on June 30, 2005 at 06:58 PM in Tabulas | 2 Comments

Related Entries

Want to comment with Tabulas?. Please login.

Comment posted on July 2nd, 2005 at 02:01 AM
If you are planning on being multilingual, then you should consider the implications of forcing Unicode down Japanese throats. The Han unification ( http://tclab.kaist.ac.kr/~otfried/Mule/unihan.html ) looks messy.
Comment posted on July 2nd, 2005 at 03:09 PM
Well, if I go the UTF-8 route, I should be OK with everything. Another option is to remember what encoding I get with large chunks of text (entry/content pages), and then just run the thing through a parser if the end-user changes their encoding.

But, realistically, no one's gonna be changing their encoding more than once, so I should be ok... but I still need to figure out how to deal with everything (I'm almost sure that UTF-8 maybe the only way to go here)