[thesite] blockfactor revisted

Daniel J. Cody djc at starkmedia.com
Thu Oct 25 23:58:17 CDT 2001


.jeff wrote:


> i'd guess char_len() or something like that that would count the number of
> characters.


not familiar with that technique
 
> i think your open source photon laser is mis-aimed.  more likely than not
> the 32K limit has something to do with the communication layer between cf
> server and the db server, something which macrollaire has no control over.


ok.

seeing that MM(allaire) wrote the software that communicates between CF 
and the DB server(more speficially, the DB driver), i'd say they're very 
much in control of that issue. assuming CF is writtin in C, they simply 
wrote some lame ass function that can only handle a 32K buffer( a 
longint in C) at a time between their app and the driver. drivers, like 
the OCI8 driver, can handle *huge* amounts of data.

thats why the same query in PHP is faster than CF using the same driver.
thats why the same query in JSP is faster than CF using the same driver.

dont try to blame it on open source. the fact is that CF isn't the best 
in ths area because they (maybe) had 20 people(which would be a huge 
amount of specific developers) working on this problem. the truth is, 
they're probably not the smartest 20 people in the industry in this 
partiuclar practice. far from it. i know you love cf and so do i. lets 
not blame problems of performance on us though. the fact is MM isn't the 
place where cutting edge, hardcore, performance code is written.

that *is* their fault, and the reason a lot of people question CF in the 
first place.

sorry i'm trashing them and shit, but a $5000 product should be about 
more than RAD. unfortunatly for MM, more people than me are thinking 
this way lately, hence the popularity of languages like PHP and JSP.

.djc.






More information about the thesite mailing list