Once the user enters their information into a form on that page, our script would take over and submit the information to our server, at a different domain, via ajax. Our server would return the result of the validation of the information.

There is only one very big problem with this approach – browsers currently consider such a cross-domain ajax post a vulnerability. Browsers will not allow an ajax post to a domain that is different than the domain currently in the browser location. Check out the same origin policy.

I won’t dive into an argument for or against allowing cross-site ajax posting here, instead I’ll just provide the mechanism we used to perform this post without using the standard ajax method, eg. XMLHttpRequest.

At a high level, what we do is this:

• Capture user input on the client using javascript.
• Construct a url containing our user input validation resource, a random identifier, and the query parameters from the user input.
• Create a new “<script>” DOM node with a src reference to the constructed url. This results in an implicit request being sent by the browser for the javascript source.
• Our server handles the request, validates the input, and returns pure javascript as a response. Embedded in this javascript is a function which will call javascript already sitting on the browser once it is loaded.
• Once the javascript response is loaded into the DOM by the browser, the script is executed – performing whatever processing is necessary at the client.

Below is the main script. This code handles user input and submits the input asynchronously by inserting a new script node into the DOM. This forces the browser to request the src for that script node. On the server, that request is handled, the input is validated, and javascript is returned back to the client which just executes the ‘responseCallback’ method.

Main Script:

Below is the server code that handles the request, validates the input from the client, and returns a response that is basically a javascript file. That javascript file contains a script that calls the ‘responseCallback’ method above with the result of the input validation.

Server Code (Java):

I both tarred up the repository directory and used ‘svnadmin dump’ to backup the repository. I scp’ed both files to another server.

Upon recovery I wanted to use the dump but the dump file became corrupted.

Luckily, there’s an easy way to recover from a repository file backup. I just untarred the repository and used

It worked great!

This was actually displaying in the blog posts, which is bad for a number of reasons, including security.

You can turn off php error logging by finding your wp-config.php file, usually in your WordPress document root, and adding the following lines:

My girlfriend and I arrived in Las Vegas ready for a great trip. Two weeks hiking around the grand circle. We had our rental car arranged and the price set.

We waited in line at Dollar for an hour while one clerk serviced the 40 people in front of us. This was starting off well. But hey, it was the cheapest.

We finally get up to the desk. I know the drill: just answer no, no, no to everything. So I did.

clerk: "Do you want insurance?"
me:    "No thanks, we don't need insurance. We have our own insurance."
clerk: "Do you want daily coverage?"
me:    "No. AAA will cover us for that."
clerk: "Do you want...?"
me:    "No. No. No."
clerk: "Ok, initial here, here, here."

I’m in a rush to get out of there at this point so I initial and we’re on our way.

The trip is great, we see lots of shit, I’m happy.

We get back to the northeast and find that Dollar charged us $340 more than we bargained for. They hit us up for an “LDW” charge, or Loss Damage Waiver. What the fuck?

I call them several times, dispute the charges online, send faxes with our original confirmation and price quote. I get the same bullshit: “Sorry. You initialed. We won’t reverse the charges.”

Ok, I did initial. I was in a rush. I didn’t read the fine print. Shame on me. But I did tell the clerk at least three times I didn’t want that garbage insurance or LDW and I have a confirmation letter from Dollar stating we would be charged $340 less!

I guess I’ll have to take this one because, well, Dollar scammed me legitimately. I initialed. Nonetheless it’s still a scam. I feel insulted and pissed.

My takeaways from this experience:

• Dollar Rent a Car will try to rip you off with LDW charges – buyer/renter beware definitely applies
• The customer is not so important as a person but is very important as a wallet
• The immediate bottom dollar is much more important than long term revenue or return revenue

Dollar has lost my business, my friends’ business, and hopefully your, the reader’s, business.

Check these links for other customer experiences or search google for “dollar rent a car scam”:
http://www.complaintsboard.com/bycompany/dollar-rent-a-car-a59315.html
http://www.rateitall.com/i-7022-dollar-rent-a-car.aspx
http://www.consumeraffairs.com/travel/dollar_rates.html
http://www.complaintsboard.com/complaints/dollar-rent-a-car-c265415.html
http://www.planetfeedback.com/dollar+rent+a+car+systems+inc/billing/payment/dollar+rent-a-car+insurance+cost+fraud/179207

Finally I stumbled upon this post, which fixed my problem.

It turned out that an nvidia driver update broke RDP. Updating the driver to the latest WHQL fixed the problem.

I love windows updates, they usually break more than they fix.

The first step was to generate certificates. This is easily done with openssl. I already have a key and scripts setup to generate cert requests with all necessary info filled in. The script looks something like this:

Take the output from that certificate request and provide it to your favorite signer to get a signed certificate. Take the key you used to generate the request and the signed certificate and put it somewhere on your server, say /etc/ssl/crt. Also make sure to put the cacerts bundle, or signing certificate chain, in that directory (or any other for that matter).

Next step is to configure sendmail. The following are the changes I needed to make to my sendmail.mc file under /etc/mail:

Most of that config points sendmail to your keys and certificates to be used for server and client mode. The line

define(`confAUTH_OPTIONS’, `A p y’)dnl

tells sendmail to perform smtp authentication after TLS negotiation has completed. The line

define(`confTLS_SRV_OPTIONS’, `V’)dnl

tells sendmail to skip requests for clients’ certificates.

I would like to thank this site and this site for that helpful information.

Next, recompile the config file and restart sendmail with

You can test your server using openssl:

You should see “Verify return code: 0 (ok)” near the end of the output. Type “quit” to end the communication.

To test that sendmail will communicate properly as a client with another server, you can use the great site test.smtp.org.

For instance, in the shot below, I have Sqeezecenter looking for music under the path s:\completed (a mapped network disk). I also have playlists under s:\playlists.

At first, I had my playlists referencing the UNC pathname to my music, so that the mapped drive letter wouldn’t be necessary (as it might be different on different computers). Squeezecenter scanned my playlists and found the file references, but because the pathname was different the software assumed the files were different and indexed them as different files. This resulted in duplicates in my library for all music I had in playlists. To solve this, I just changed the playlist paths so that they matched what I had set in Squeezecenter. A new scan of the library and playlists and Squeezecenter correctly recognized the files as being the same and didn’t produce duplicates.

So in your playlist files, change this:

\\storageserver\Completed\Hard Rock\White Zombie\La Sexorcisto – Devil Music Vol. 1\05 White Zombie – Soul-Crusher.mp3

To this:

s:\Completed\Hard Rock\White Zombie\La Sexorcisto – Devil Music Vol. 1\05 White Zombie – Soul-Crusher.mp3

As an aside, it would be nice to set a UNC network path within Squeezecenter, but it doesn’t currently allow it. I believe an older version did allow you to set your library as a UNC path on the network.

After some traffic inspection and a look at buglist.cgi, I found that new code was added to “clean” advanced bug queries. This code cleans the query, then performs a redirect to the new clean query using the correct bugzilla hostname but using the local webserver port. Because I was proxying traffic to my actual bugzilla server through another server on a different port, this redirect was invalid.

I fixed this by commenting out the following code in buglist.cgi:

I’ll try to actually fix the redirect tomorrow to take advantage of the cleanup.