I think I found a bug in Error Tracker 1.06, a plugin for Mint.  I came across this one because even though I had taken measures to redirect a certain incorrect URL that was apparently getting a lot of hits, the hit count on that URL continued to climb.  I ran my own queries against the database to confirm my suspicion that the numbers were wrong.

My findings and solution are found in comments 17, 18, and 19 on the Error Tracker page at XHTMLed.com.

Here’s the current bug list for Mango that I’ll be working on over the next week or two. After this list is complete, the initial release will go online. Nightly backups of data will be required just in case I miss something which causes the database to become corrupted.

registration
double what do i do? fixed
auto focus fixed
registration is stupid if it works fixed
after reg, auto login the user fixed

login
pressing enter is no work in form fixed
show me that i logged in fixed

logout
show me that i logged out fixed

general
New Document link should be removed
feature request: search next version
powered by link goes nowhere

pages
add spacing to title fixed

edit page
upload ten files, click save. expectation: files are added to revision. actually: files are saved to document, must be manually attached to revision. fixed
manually attaching files doesn’t seem to work fixed

all pages
it would be nice to allow chronological sorting as well next version

help
seek an anchor based on action (index/add/view) next version
pages that need help fixed

• all pages
• become an author

profile
redirect this page to the edit page. who the hell needs to see this? move the updated and joined dates over to the edit page as information. fixed

edit profile
remove roles from page if not admin fixed
double what do i do fixed
Submit button should say “Save” fixed
save should save, then go back to editing page with a message that things were saved fixed
feature request: add button for “pages i’ve edited” next version

administration
lock down to admin permissions fixed

new document
creating a new page (at least by going to the url) doesn’t seem to work fixed

I was right. If the server sends a response and the ‘Content-Type’ attribute is ‘text/xml’, you must send at least a root node so the xml parser that is built into Mozilla’s XMLHTTPRequest object doesn’t cak and give you a ‘no element found’ Javascript error.

This could be an issue for those of us who are trying to do the REST thing and send back proper status codes depending on the request method and actions taken by the server. A 204 (Updated) or 205 (Deleted) response cannot, by definition, have any content body, so sending a root node is not an option to get around this error. The server needs to determine if the response is going to have an xml content body and assign the Content-Type on the fly.

I think I figured it out. I think that the issue crops up if your server sends a response header of ‘Content-Type’ set to ‘text/xml’ and you send no content. The XMLHTTPRequest object then says ‘oh it’s xml? fine - i’ll just PARSE THIS and store it in the reponseXML property.’ At that point, the XML Parser probably bombs and nothing catches it. But anyway that’s just a hunch and I’m only blogging about it because I’m bored. I’ll have test results available later.

Firebug 0.4.1 and 0.4.0 suffer from a very peculiar bug. If you create an object with a property called ‘-URL’, you can not use console.log to see the value of the property using the console.

Steps to reproduce:

In the console, type

x = {'-URL' : 'foo.pdf', '-Size' : 100};

Now type

x

Now press enter and click on the [Object object] item which shows up in the console. You will see the ‘-Size’ property, but not the ‘-URL’ property. Next, enter the following into the console and press Enter:

x['-URL']

You should now see “foo.pdf” get output to the console. So the value is in Javascript memory, but Firebug won’t show it to you. On a related note, you also cannot view methods of a custom object which are defined using the same means as described above … most of the time. Sometimes it works, so I can’t say for sure what the steps are to reproduce this behavior every time. Why am I blogging about this? Hopefully it will save some other Javascript developer a few hours. Credit goes to my friend, Matthew Metnetsky, for showing me the issue.