[thelist] Slowing Down a Connection

Ron Thigpen rthigpen at nc.rr.com
Fri Mar 30 08:40:02 CST 2001


The document weight tool should focus on counting the number of bits transmitted to
the client, and thereby estimate the time for those bits to travel
server-to-client.  

The best way to measure this is by rendering the page through the webserver, and
then 'weighing' the result. This would work much like the 'File: Open from the Web'
feature.  Writing code that follows links willy-nilly up and down the file system is
just reimplementing what the webserver/CFserver combo are already doing for you.  
If the document weight tool were going to be updated this is how I think it should
work.

BTW, you can get some of this data from the browsers. There is a web developer
add-on for IE that weighs the images for you, adds them up and does the math for
download times at various connection speeds.  Once installed it's right there in the
context menu (right-click...)  Oddly, it doesn't add the weight of the HTML to the
image weight.  
Perhaps some enterprising soul will open the hood on this and add that feature
(*Hint*).  It appears to be done in simple client-side scripting.

In my opinion, the document weight tool should not try to be a code optimization
tool for server-side processing.  There are already plenty of tools for doing that
(debug info, query exec times, page exec times per include, sql query optimizers,
etc.)

--Ron Thigpen




martin.p.burns at uk.pwcglobal.com wrote:
> 
> Memo from Martin P Burns of PricewaterhouseCoopers
> 
> -------------------- Start of message text --------------------
> 
> Evaluating basic includes would be good, but not the final solution.
> You'd also need to evaluate exec type includes, not only for data
> transmission, but also for execution time. Oh, and XSSI commands too.
> 
> They're very fast, but if you've got lots of conditional XSSI, then you
> could seriously mislead a simple character counter (ie the file has
> 30k of code, but only 10k goes to the browser)
>




More information about the thelist mailing list