Re: [CFCDev] CFINVOKE vs. Evaluate

[Brian Rinaldi]

I think you are painting with a pretty broad brush here. I think the point is that it is preferable to avoid, but actually has its limited uses. I also think that saying that "it demonstrates a lack of understanding of the features of the language " is mistaken as well. The use of evaluate alone does not, in fact, demonstrate this...and by coloring everything black and white you lose alot of detail. I cannot think of a programming rule to which there doesn't exist exceptions, and the reasoning that has been offered thus far does not lead to such grandiose assertions. I am, by the way, saying this as someone who has managed to avoid evaluate ;)

- Brian Rinaldi
blog: http://www.remotesynthesis.com/blog
ColdFusion Open Source List: http://www.remotesynthesis.com/cfopensourcelist
Boston CFUG - http://www.bostoncfug.org

---

From: "Brian Kotek" <[EMAIL PROTECTED]>
Sent: Thursday, May 18, 2006 2:19 PM
To: CFCDev@cfczone.org
Subject: Re: [CFCDev] CFINVOKE vs. Evaluate

The point is that in virtually all cases:

1. evaluate() is slower
2. there is no reason to use it
3. it is more difficult to read and understand
4. it demonstrates a lack of understanding of the features of the language

On 5/18/06, Roland Collins <[EMAIL PROTECTED]> wrote:

Resent without the "SPAM" flags ;)

 

I've actually posted metrics on this before – and you're pretty much right.  The performance penalties of using evaluate are trivial at best, and often times non-existent.  The metrics below demonstrate that, given server conditions, the evaluate code can even perform faster than inline code  Anyway, it's *very* clear from the results that the overhead of evaluate is largely over-hyped.  Also, attached are the two test files I used to generate the metrics.

 

Simple _expression_ - 10,000 Iterations
	Inline	Evaluate
Run 1	32	62
Run 2	31	78
Run 3	16	62
Run 4	31	79
Run 5	31	63
Average	28.2	68.8
Simple _expression_ - 100,000 Iterations
	Inline	Evaluate
Run 1	328	437
Run 2	312	828
Run 3	297	485
Run 4	313	828
Run 5	312	578
Average	312.4	631.2
Object Invocation - 1000 Iterations
	Inline	Evaluate
Run 1	1830	2190
Run 2	1328	1391
Run 3	1437	1078
Run 4	1313	1843
Run 5	1843	1688
Average	1550.2	1638
Object Invocation - 10000 Iterations
	Inline	Evaluate
Run 1	14375	11672
Run 2	14250	20187
Run 3	17140	20828
Run 4	18328	11735
Run 5	20453	20125
Average	16909.2	16909.4

 

Roland

 

---

From: [EMAIL PROTECTED] [mailto: [EMAIL PROTECTED]] On Behalf Of Brian Rinaldi
Sent: Thursday, May 18, 2006 11:05 AM
To: CFCDev@cfczone.org
Subject: SPAM: Re: [CFCDev] CFINVOKE vs. Evaluate

 

I only recently started watching this thread, but this seems to me that the "avoid evaluate()" rule generally doesn't have much documented justification (though I am not arguing that it isn't valid - just that I have rarely seen much documentation on the differences - if I missed some earlier in this thread I apologize in advance). The ColdFusion MX Coding Guidlines only says the following under Performance "Don'ts":

"Don't use evaluate() - Avoid evaluate() unless there is no other way to write your code (and there is almost always another way to write your code)."

A Google search only comes up with the following posts (the first being recent and the second a couple years old):

http://cfprimer.blogspot.com/2005/03/avoid-evaluate.html
http://www.sitepoint.com/blogs/2004/02/19/running-performance-metrics-on-coldfusion-functions-and-operators/

An MXNA search comes up with some more recent posts:

http://www.succor.co.uk/index.cfm/2006/4/13/Avoid-evaluate-always
http://www.blogoffusion.com/index.cfm/2006/4/18/Structure-style-evaluation-rocks

However, nobody seems to offer specific metrics (except for one that came to the conclusion that the differences were minimal). Let me be clear, I am not arguing that this isn't a completely valid best practice - I am sure at some point it was based upon some documented metrics, but it would be nice to see them.

- Brian Rinaldi
blog: http://www.remotesynthesis.com/blog
ColdFusion Open Source List: http://www.remotesynthesis.com/cfopensourcelist
Boston CFUG - http://www.bostoncfug.org

---

From: "Nick Tong - TalkWebSolutions.co.uk" < [EMAIL PROTECTED]>
Sent: Thursday, May 18, 2006 10:49 AM
To: CFCDev@cfczone.org
Subject: Re: [CFCDev] CFINVOKE vs. Evaluate

On 18/05/06, Peter Bell wrote:
> So, would cf invoke actually be any faster?

I think this can depend on if you have initiated the object into memory i.e.


in this case then i believe it is faster

--
Nick Tong

web: http://talkwebsolutions.co.uk
blog: http://succor.co.uk
shorturls: http://wapurl.co.uk


----------------------------------------------------------
You are subscribed to cfcdev. To unsubscribe, send an email to cfcdev@cfczone.org with the words 'unsubscribe cfcdev' as the subject of the email.

CFCDev is run by CFCZone (www.cfczone.org) and supported by CFXHosting (www.cfxhosting.com).

An archive of the CFCDev list is available at www.mail-archive.com/cfcdev@cfczone.org

----------------------------------------------------------
You are subscribed to cfcdev. To unsubscribe, send an email to cfcdev@cfczone.org with the words 'unsubscribe cfcdev' as the subject of the email.

CFCDev is run by CFCZone (www.cfczone.org) and supported by CFXHosting ( www.cfxhosting.com).

An archive of the CFCDev list is available at www.mail-archive.com/cfcdev@cfczone.org

----------------------------------------------------------
You are subscribed to cfcdev. To unsubscribe, send an email to cfcdev@cfczone.org with the words 'unsubscribe cfcdev' as the subject of the email.

CFCDev is run by CFCZone (www.cfczone.org) and supported by CFXHosting (www.cfxhosting.com).

An archive of the CFCDev list is available at www.mail-archive.com/cfcdev@cfczone.org