Updates for 02/26/2011

4 posts / 0 new
Last post
puravida's picture
puravida
Jedi Warrior
Offline
Joined: 09/01/2007
Visit puravida's Website

We are testing some changes to the way that encoded URLs are handled.

APPLIES ONLY TO INSIDE PAGE REQUESTS

This will allow for wider support of non-default encoding strings (such as %92, a tick mark) that require a specific, defined encoding. Therefore, for URL input, we will be supporting ISO-8559-1 for maximum Western character-based support, unless we find that another character set works better. However, our XML feed output will still remain in UTF-8 for maximum international compatibility.

Testing requires changes to the generator code and the delivery code. This should not affect any non-inside page requests. However, if you notice anything out of the ordinary (assuming that your request worked prior), then please let us know.

We do not foresee any impact to services as a result of these changes. We will post an update to this thread once the changes have been completed, tested, and implemented LIVE.

puravida's picture
puravida
Jedi Warrior
Offline
Joined: 09/01/2007
Visit puravida's Website

These code changes are LIVE now.

After more than 10 hours of research and testing, I have discovered that our system was already UTF-8 compliant for all requests (maximum international compatibility) and also UTF-8 compliant for all text output. The problem with the %92 support is that it is not UTF-8 compliant but requires ISO-8859 or ISO-8559 character sets. I got the system working with that version before discovering during testing that our rendering engine (loosely based on a common browser platform) requires the UTF-8 compliant representation of the tick mark in order to successfully browse to the page URL using it. Therefore, I have made the following changes (slightly different from the original post plans):

  1. Forced UTF-8 character set in our delivery code
  2. Forced UTF-8 character set in our render engine capture code
  3. Added better handling for scheme when dealing with encoded URLs
  4. Added better handling for single quotes in URLs*

*This means that, technically, %92 is not supported by UTF-8 and won't be supported by us. However, it also means that you can send a UTF-8 compliant single quote (%27) or the single quote itself (now that we've added better handling for that character).

All of these changes have been thoroughly tested and run through more than 5,000 new requests without issue (that we noticed). If you believe that you have found anything not working correctly in regards to this change, please let us know.

puravida's picture
puravida
Jedi Warrior
Offline
Joined: 09/01/2007
Visit puravida's Website

BUG FIX: Optimization routine for Public-Use generators was not generating properly

Details: In the midst of our testing, I realized that there was a typo in the optimization routine governing the Public-Use (i.e. free user) generators. Under heavy load, it was possible for the queue to back up significantly or stop altogether. Since we have many more generators than needed, we haven't experienced a load heavy enough to bring this bug to light before now. However, it is fixed now so it won't be an issue, if we do reach a point of heavy load on the generators.

puravida's picture
puravida
Jedi Warrior
Offline
Joined: 09/01/2007
Visit puravida's Website

Enhancement: I also noticed that the thumbnail history was breaking (wider than the screen) on very long encoded URLs.

Details: I decided to account for this by showing all encoded URLs in a fixed width text box rather than truncating it (the way non-encoded URLs are handled). The reason I chose this was to allow the full encoded URL to be copy/pasted for easier testing.

ShrinkTheWeb® (About STW) is another innovation by Neosys Consulting
Contact Us | PagePix Benefits | Learn More | STW Forums | Our Partners | Privacy Policy | Terms of Use

Announcing Javvy, the best crypto exchange and wallet solution (coming soon!)

©2018 ShrinkTheWeb. All rights reserved. ShrinkTheWeb is a registered trademark of ShrinkTheWeb.