I'm grateful for and to humanity's heroes, and the people who achieve.

MSNBC is a news Web site. There how does that look? Is it OK? Golly! I hope so!

Yea! It looks like it's working. Is it really?

Test of linking shortcut. My previous tests directly using HTML seemed to work. The above link uses the WoC shortcut.

We use similar code at Discarded Lies. However, the requirement that MT rebuild the recent comments include was just too weird as an experience for the user. A reader would post a comment, the thread would reload, and MT would still be rebuilding the include in the background, so the recent comments sidebar would not include their newest comment. Then, the next time they reloaded, it would be there. Winds of Change doesn't have this problem because you're not showing the recent comments in the sidebar of an individual entry, so by the time a reader returns to the front page, their comment will already have appeared in "recent comments".

Instead, we went to dynamic php pulling the recent comments live from the database, to avoid the rebuild lag. Here's some code, simplified from what we actually use:

<? //grab 20 newest comments from the database
$result = mysql_query("SELECT * FROM `mt_comment` ORDER BY `comment_created_on` DESC LIMIT 0, 20");
//was there an error?
if(!($result)) {
print "comment id select error > ";
}
//start unordered list
echo "<ul>";
//loop through the 20 comments we grabbed
while ($row=mysql_fetch_array($result)) {
//author is?
$cauth=$row["comment_author"];
//comment text is?
$cbody=$row["comment_text"];
//put spaces around < and >, which puts spaces after html tags that might be crowded together. we do this because we're not going to strip all tags; instead, some tags, we want to indicate their presence in the recent comment preview
$cbody = preg_replace('/</',' <',$cbody);
$cbody = preg_replace('/>/','> ',$cbody);
//make bolded words have stars around them
$cbody = preg_replace('/<(\/|)b>/', '*', $cbody);
//make italicised words have underscores around them
$cbody = preg_replace('/<(\/|)i>/', '_', $cbody);
//strip all other html tags, then decode all html entities
$cbody= html_entity_decode(strip_tags($cbody));
//replace linebreaks and tabs with spaces
$cbody = preg_replace('/[\n\r\t]/',' ',$cbody);
//eliminate excessive spaces
$cbody= preg_replace('/ /',' ',$cbody);
//reencode html entities
$cbody=htmlentities($cbody);
//get comment's entry id, so we can build a link to it
$cmtentryid=$row["comment_entry_id"];
$entryresult=mysql_query("SELECT `entry_title`, `entry_created_on` FROM `mt_entry` WHERE `entry_id` = '$cmtentryid'");
if(!($result)){print "entry id select error > ";}
$entryresultrow=mysql_fetch_array($entryresult);
$ecreateon=$entryresultrow["entry_created_on"];
$etitle=$entryresultrow["entry_title"];
$dateresult = mysql_query("SELECT UNIX_TIMESTAMP('$ecreateon')");
$datetime = mysql_result($dateresult, 0);
//having grabbed the entry's date and title, build out the comment's url
$curl="http://discardedlies.com/entries/" . date("Y/m/", $datetime) . dirify($etitle) . ".php#c" . $row["comment_id"];
//we use the title attribute to show an even longer excerpt of the comment in a tooltip when the reader hovers on the recent comment. IE and Opera users can see this, although Firefox pathetically truncates title to not much longer than the excerpt itself. first we grab and sanitize the entry title:
$etitle=htmlentities($etitle);
//then we build the whole thing.
echo "<li title=\"entry: " . $etitle . "; comment: " . substr($cbody , 0, 300) . "\"><a href=\"" . $curl . "\">" . $cauth . "</a>:<br />" . ($row["comment_visible"]?substr($cbody, 0, 30):"held for moderation") ."</li>";
}//end of loop, now do this for the next recent comment, till you've built all 20
echo "</ul>"; ?>//end unordered list

I'm puzzled why Joe's technique is working. As near as I can tell, remove_html does what it says - it removes HTML. But a URL is not HTML, it's a URL. And URLs are the long strings that cause these problems.

I manage a site that has been using a technique almost identical to the one Joe describes for several years. It works nicely in every way except for the fact that long URLs in comments still cause problems in sidebars for IE users.

I'm looking for a solution that will actually remove URLs. Any suggestions?

For whatever reason (note test comments above, which worked), it no longer works. It doesn't matter whether the URLs are hard-coded as HTML in the comment box, or attached via Textile coding that then gets converted. In both cases, not only does the URL show up but so does the HTML coding.

This says to me that we have a simple problem called "the Movable Type tag is badly coded and doesn't work". A well coded solution would indeed strip all HTML coding formats and items enclose in the > and < out of the text, which would erase the URL strings as well as the HTML tags while leaving the transformed text untouched.

The cure sits in Six Apart. Unless there's a way to fix their bad coding through a plug-in that wouldn't have other undesirable effects.

I think the important distinction is that when a URL is in an href, it's HTML. But when a user just pastes in a URL and it's not part of a link, it's just a string like any other.

What complicates things, I think, is the "Auto-link URLs in comments" setting. Yes, it links the URL, but it links it to itself, i.e. one URL in an href and one as a string. I'm betting that if that option allowed MT to just create a link that said [link] or similar, the problem might go away. Not sure.

I've posted this question on the MT support forum as well [link] :)