Exactly. Why would we want something posted that will mess up other posts or the entire showthread.php page?

Why would we want something posted that will mess up other posts or the entire showthread.php page?

We absolutely wouldn't.

I wonder, however, about two things:

1. We get caught in a user-controlled loop of no submission (admittedly, a really easy one to escape).
2. Before we can run tidy on it, we have to do the expansion from quickedit. I wonder if there's a risk of a combo of bad HTML entered by the users, and expanded quickedits that would give a tidy error, but not be easy to eyeball as a problem for the user -- see, we're not presenting the HTML that would be posted to the user to be fixed.

Okay, now I'm obsessing about #2.

1 isn't a problem because the user can always submit the Tidy'ed version as presented. I'll have to think on 2.

Once we get this figured out, I think we should start Tidying content. We could start with not allowing a post to be submitted if it contains two or more LOL's.

Okay, now I'm obsessing about #2.

Wow, it's like you are in a controlled loop and won't submit.

t ducks

Before we can run tidy on it, we have to do the expansion from quickedit.

Hmm. Do we? Unless something goes blewwy, can't we know that the quickedit expansion is going to be kosher?

If the resubmitted post still has html problems, the process begins anew.

The theory being that crap is never allowed through?

While this is nice, the thing I worry about is causing the user experience to be potentially frustrating. Sometimes we've had posters need to ask what they did that caused the problem, and if they were stuck inside a no-post loop, they might not be able to ask.

Exactly. Why would we want something posted that will mess up other posts or the entire showthread.php page?

Well, for me there's a difference between board-breaking errors and something along the lines of unclosed italics tags. One is significantly problematic and should probably be contained if possible, the other is merely inconvenient for as long as it takes us to get to it (currently, of course, and I realize we're trying to improve on this so we don't have to get to it, since they're almost always rote errors). For the former, I think it's fine to potentially frustrate the user a tad, so long as it preserves board integrity. But for the latter, I don't think it's enough of a problem to stick the user in a potentially recursive loop.

Unless something goes blewwy, can't we know that the quickedit expansion is going to be kosher?

I know if we search the archive, we'll find posts from me that look like:

Did you really not like
i The Incredibles</i>?

Not so many, nor recently, until I thought it through and realised how lazy and dumb a way it was to post.

Just a thought about how the two can tangle.

Ah. This is true. Okay.

Sometimes we've had posters need to ask what they did that caused the problem, and if they were stuck inside a no-post loop, they might not be able to ask.

They accept the Tidy'ed version and then they ask.

Unless something goes blewwy, can't we know that the quickedit expansion is going to be kosher?

Here's an example of a post with bad HTML we've run into a couple of times recently:

<font size="-3">
>blah blah blah</font>

Because the opening font tag is outside the quikedit and the closing font tag is inside the quikedit, on some browsers, the small font continues to the next post.

The same TidyPlus operations can be performed on both ends.

Flow:

TidyPlus is run on a new post. If we get an error, the preview page is presented (else, save), displaying how the TidyPlus will alter the display of their typing. In the edit box, we display what the poster typed. They may alter or accept. If they accept, the save (of what they typed) goes forward even if errors are still present. If they alter, we go back to the top of the stack.

On display, we run TidyPlus, which protects the board, if not the sense of the post.

But how does TidyPlus detect post-expansion errors, yet present a pre-expansion post for the poster to correct?