[thelist] stuck using cf4.5? upgrade to cfmx - no to PHP

Jeff Howden jeff at jeffhowden.com
Thu Jul 17 04:28:24 CDT 2003


> From: Andrew
> But are we really talking about building the car?  Or
> just driving it!

neither, the car analogy sucks, unless we're talking about one car that's a
left-hand drive, front-wheel drive, automatic transmission, with
power-everything (cf) and the other is a right-hand drive, rear-wheel drive,
manual transmission, with hand crank windows and pull-knob locks (php).
then, and only then, are the differences significant enough to warrant the
kind of learning curve to switch between cars as a developer would have
switching from one language he/she is comfortable with to another that
he/she isn't familiar with.

> Firstly the cost is significant between the two. So the
> Win car is expensive and the Linux Car is free.  But you
> still need to learn to drive regardless of which one you
> are going to drive (you may want to drive both ;))

i think you believe the cost is significant because you don't realize the
investment you've already made in one.  if you're starting from square one,
the long term costs are insignificant.  if you're already grounded in one,
there's precious little incentive to make such an investment.

> The point to make is that if you going to drive a 'car'
> and have to learn anyway it would make sense to get the
> free one, that coincidentally has features just as good
> as the Win car and as it also happens drives better!

that's a matter that's certainly debatable.  i personally find any supposed
claim to better performance with linux to be overshadowed by the increase in
management time when something goes wrong or something needs

> Performance comparisons:
> most recent) http://www.kegel.com/nt-linux-benchmarks.html
> older)
> http://www.csu.edu.au/special/auugwww96/proceedings/elson/elson.html

neither of the findings at the links above support your notion that linux is
better than win2k.  in fact, the first one shows with some very stringent
benchmarking that win2k3 is over 1.5 times faster than linux' red hat
advanced server 2.1.  there are the typical quibblings about tests being
skewed because the benchmarking company is "in bed with microsoft", certain
features on the linux install "may not have been enabled or configured
properly", etc.  however, none of this smokescreen to avoid the truth can
make up for the vast difference in numbers.  regardless of how much faster,
win2k3 is simply faster.

> The fact that the average bod wouldn't know how to
> change the oil of a car let alone build one is just
> another reason not to buy the Win Car!  Remember, it
> was built in secret and the remains truly hidden under
> the hood whereas the Linux car can be opened up and
> checked out!  With, as it happens mechanics available
> globally and almost at anytime of the day, can't say
> that the Win Car mechanics.

actually, i could give a sh!t about how the car actually works.  i just want
it to start every morning when it's time to go to work.  i'm not in the
business of maintaining the car.  i'm in the business of accessorizing it
(web applications, we sites, etc.).  the car is just a means to an end, a
delivery mechanism.

i don't want to know how to change the oil, adjust the motor speed for the
electric windows, change the timing, adjust my fuel mixture, etc.  i'm not
interested in being able to dig into the code that makes the file system
work.  i don't have time to even familiarize myself with the language it's
written in, let alone tweak it, write down what i did, benchmark it against
previous configs, and then be sure to make those tweaks every time i apply
an upgrade or recompile the kernel.

> Interestingly PHP and CF can both be features of either
> the Win Car or the Linux Car, so how do they compare?
> This makes an interesting read:
> http://php.weblogs.com/php_vs_cold_fusion

not interesting at all actually.  it's really quite an out-dated bore.

here's some things i noticed right off the bat.

written in late 2000.  it's nearly 3 years old now.

platform support:
cfmx will run on most anything you can install a compatible java servlet
engine on.

cf is *not* built solely for display code.  it supports user-defined
functions.  i'll agree that the operator syntax is not what many coders with
scripting-like experience are used to, but it's surely not something to
knock it for.  it does support most any of the operators you might wish to
use.  it's scripting language has actually gotten quite robust.  in fact,
developers now have the choice to use actionscript, very much like
ecmascript, within their cf templates.  in cfmx there is also the ability to
create and consume webservices.  can php even do that yet without lots of
wrangling with line after line of messy code?  cfmx also has built-in xml
handling functionality.  read in an xml doc and convert it to native
data-types in memory and vice versa.  in cfmx, components offer cf
developers oo-like development environment, akin to the oo support in php.

database support:
drivers are included for jdbc and type 4 drivers.  you can connect to most
any kind of database that'll accept external connections.  you can even
configure a text file as a datasource.

regex support:
i'll agree that it's probably not as full-featured as perl might be.
however, i've yet to run into a situation where it wouldn't do what i needed
it to do.  i suppose some of the most geeky of the regex geeks might find
some restrictions they just "can't live without", but i don't see those days
in my own future.

not only is error-handling awesome, the error-reporting is phenomenal as
well.  i've yet to find a middleware product that comes even close to cf in
this department.  i can only begin to imagine how much time this alone has
saved me.

in my experience, cf on win2k can be very solid.  in fact, evolt.org is
hosted on a win2k box running cfmx.  it's not a dedicated box.  there are a
number of other clients contributing to the load on the server.  a couple of
days ago evolt.org got slashdotted because ppk's article got posted to
slashdot.  the box withstood some serious amounts of traffic without so much
as a change in "heart rate".  here's some of the numbers for evolt.org for
the 24 hour period i ran a report for:

Duration (hh:mm:ss)        23:59:50 (1 Day) 7/14/2003 (gmt)
Total Hits                 1,011,690
Total Cached Hits             66,999
Unique Visitors               66,409
Visitor Sessions              87,075
Average Visitors Per Hour     2,767.36
MB Transferred                8,431.65

i'm sure php can post some numbers like that as well.

my point?  it doesn't really matter which one is better, unless you don't
have experience with either and then it pays to get informed and pick the
one that makes the most sense for you.  however, if you're already well down
the path you've chosen, then the cost to move from one to another is
significant and most likely extremely foolish.


Jeff Howden - Web Application Specialist
Résumé - http://jeffhowden.com/about/resume/
Code Library - http://evolt.jeffhowden.com/jeff/code/

More information about the thelist mailing list