View Single Post
  #5  
Old 10-14-2003, 11:40 AM
Merth
Dragon
 
Join Date: May 2003
Location: Seattle, WA
Posts: 609
Default

Quote:
Originally Posted by Trumpcard
Image, you nailed it on the head..

One of the big problems in enterprise level applications is that once your architecture is down, changes are very hard and very expensive to make and test
While this is true in companies I have worked for, I disagree with regard to EQ. That's why I frown on their recent attempt to cut down bandwidth costs.

EQ recently underwent major changes to the underlying protocol. It did not do this upon arrival of an expansion, it was between expansions. In fact, there were two major protocol changes between December 2002 and April 2003 - and is the reason why EQEMu and ShowEQ were out of commission for quite a while.

Their September 9, 2003 change to items was not as huge as the compression/encryption changes, but it was still a big change - items were not only serialized as a string, but they put items inside of other items, and did away with the slot_id for these items. This was definitely an improvement to bandwidth usage, but they are missing the bigger picture on what they are trying to accomplish - and this is where some major bandwidth improvements could have been made.

This carries over into some other packets being sent as well, but my point is: they are well suited to make big changes, but they end up making the wrong changes
Reply With Quote