I think this <code></code> and <p></p> parser would be harder to implement than it seems (even keeping in mind your counter points). [...] There is also the question of how do you parse out incorrect code/pseudo code.

Does that parser have to be complete and perfect? All we want at this point is a simple check that the raw posting text contains one or more <p> and zero or more <code> or <c> tags. (Tags, not elements. We don't even have to check for the closing tags. Existing code handles missing or mis-placed tags quite well.) Unless that condition is met, the poster will see a hint that (s)he should check is posting. Implementing those checks could be done using one or two simple regexps.

Alexander

What you describe already exists (try posting something in plain text to see). Anything more will require another layer that delays/prevents a post from being submitted so - yes, it would need to be a /very/ good parser. A parser that would be much more work than would be worth the trouble considering SoPW doesn't seem to get more than a total of 25 questions per day.

The handful of weekly "evil markup violators" can easily be handled with a quick private /msg with suggestions to improve their post. If they ignore it (which does happen), ignore the post and move on. ;-)



"...the adversities born of well-placed thoughts should be considered mercies rather than misfortunes." — Don Quixote

Ignoring the bad markup isn't a complete solution. Janitors still are likely to be asked to fix it.

As to posting plain text, I've never encountered difficulty including some... as in the case of the first (untagged) para in this node, nor in the case of a single sentence of narrative reply. (alternating paras with and without tags maintains readable rendering)

Re the quality needed -- As my update, above, illustrates, coding a set of simple tests for the most egregious failures can be done even by he who frequently wears a dunce cap and without writing anything like a complete parser. If one wished to be more rigorous -- to correct ALL flaws -- then one of the html parsers should make an easy starting point.