It’s been brought to my attention that the comment button doesn’t work when people come to this site via //www.meish.org, though it’s completely fine when people come via //meish.org - probably because it thinks a spambot is auto sumbitting because it looks like the request comes from a different URL - but it isn’t different! It’s just the same with WWW in front.

I’ve looked in the comments.php file and changed things so that it sniffs the user referral string and substitutes that for the siteURL set in preferences (currently set to //meish.org), but that doesn’t seem to have any effect.

It’s driving me potty.

Can you - and this is not just a trawl for comments - please try to comment on this post and tell me what the URL in the address bar at the top of your browser is when you do? With any luck we’ll start to see a pattern and that will help to resolve it.

On the other hand, if anyone out there’s a php whizz and would like to help out briefly, please let me know.

*sigh*

Update
This problem was first spotted on the mayfly project page (linked above), but I think it must actually be sitewide, only not noticed until now.

The big difference between the mayfly page and normal comments is that when you go to //www.meish.org and then click on a comment link for any post, it takes you to //meish.org/postname etc, which means it always matches the siteURL. However, that’s not the case for people going to www.meish.org/mayfly, where the comments form is out in the open already.

Update update
Fixed!

Paul, my lovely husband, to the rescue (once again). He dove in to the templates and spotted that it was the usual horrid lack of synchonicity between PHP, Javascript and the browser that plagues Ajax apps. So, for those who want to know what the fix was, and for posterity:

He edited the parameter in the Ajax constructor that was getting the stylesheet directory path directly from Wordpress’s config as that was using meish.org as an absolute. So he hardcoded the local path, but the hostname comes from the server’s HTTP headers and so will always be right.

And now we know.

Thanks also to Simon for spotting the problem in the first place and Steve for being so quick off the mark with feedback.