I really dislike having the caption text be part of the image.

• It's bad style from a web standards perspective, because it ignores user settings for fonts, so you're breaking accessibility.
• It doesn't get indexed by search engines, nor by the forum search function, so it makes it harder to find things.
• You can't format it with the forum tools, so no italic, bold, or things like @ (very useful in captions)

It would be much better, IMO, to put a div tag around the image, set to slightly more than the width of the image for text wrapping purposes, and have the text inside that div, under the image, as normal text.

It would be much better, IMO, to put a div tag around the image, set to slightly more than the width of the image for text wrapping purposes, and have the text inside that div, under the image, as normal text.

So would you like to use any of the forum tags in the caption? If so, do you want the caption input method to remain a part of the diagram code? Right now we use sltxt2png's built in caption parsing, but if we allow for all forum tags, maybe we can do this a different way.

Ideally, the caption would just be passed back to the bbcode interpreter, yes.

The caption has already been passed through the bbcode interpreter by the time it gets to your javascript...

...

No, I'm saying the bbcode parser has already parsed any tags inside diagram code before you even get to do anything with it, and it does it in one pass.

For example, imagine we had a simpler bbcode rule that replaces: with:

Then the following: gets replaced with: or something similar (I'm pretty sure it removes the newlines, but I left them in for readability). You can see it replaced the bbcodes inside your [go] bbcode with the equivalent HTML and escaped special HTML characters like <, > and &. Smilies would also be converted to the appropriate <img> tags.

Your funkytag example would not do anything weird, because it's only a single pass. so [go1] would get converted to [funkytag] but that would be it, and viewers would just see [funkytag]. It wouldn't go into an infinite loop or anything.

My explanation may have been confusing, cause if you click on the "show diagram code" for the first example you'd see the word "caption" in italics because the [i] bbcode I put in there got converted to an <em> (as you can see in the diagram itself). So "show diagram code" is lying because it doesn't unconvert the bbcodes or anything. That's a much hairier problem. I don't know if bbcodes have a mechanism to allow certain text inside your tag to go through unmolested. My guess is that it does not because it would be really easy to create HTML injection vulnerabilities.