Coldfusion Marketshare

[fs22]

Greetings,

I have been a CFer for some time now and absolutely love it. Our company has been using it for a while now. Recently, though, some discussions have taken place within our organization about the future of CF and whether we should go to .Net or not. I am fighting for CF because I see its benefits, but my problem is that I am seen as biased since I have used and loved CF for some time. Are there any "objective" resources out there that speak to the future of CF and its benefits?

Thanks,

Clay

[ilssac]

Define "objective".

Why should your company care if any other company in the world uses ColdFusion or not?  If it works for you it works.  If it doesn't work it doesn't.

It is important to note that conversion projects of any sort are complex, expensive and have high failure rates.

Developers should not be an issue becuase I have yet to find somebody who understands any type of programming not understand CFML.

P.S.  ColdFusion now integrates with .Net.

Blue Dragon offers a .Net CFML engine.

There are other CFML engines as well.

Adobe is planning for the next version of ColdFusion in about three years.

I think the bigger question to ask, is what benifits does your organization see in .NET.

[fs22]

Thanks, Ian. I have used all of those arguments and what it has come down to is that there is an idividual in our group who loves .Net and has not taken the time to understand CF and how it is superior to .Net. I am seen as partial, so they

are wondering as to CF's future (particularly this .Net fellow) and were asking about impartial evidence.

So since I have brought all of those arguments up and struggling a touch to make my point by my partiality, I figured I would see what kind of data I can find out there.

[tclaremont]

Going with ColdFusion does not prevent you from using .net. Not sure why you would want to, but there is nothing stopping you.

The argument usually boils down to the fact that people who don't KNOW ColdFusion dont want to lobby for it because they see it as a risk to their job. If they don't know CF, but they DO know .net, it will never be as good as .net.

[ilssac]

Then let this .Net guy write .Net code.  ColdFusion can now use his code.  I'm sure there are caveots, I have never had a need to do this, but CF has had the capabilities to do for three years now.

What future do they want or need.  If their minds are made up they are made up.  It is used by nearly a million developers and growing, there are multiple CFML engines by multiple companies and organizations.  Adobe is already working on the next version.  ColdFusion is not going anywhere anytime soon.

But neither is .Net, PHP, JSP, Python, Rails or any other of the inumerable technologies people use to create web applicaitons.

So the real argument has got to be, How much is it going to cost you to abandon CF and why is this cost worth spending.

What kind of developer can't understand a different technology.  I often find these developers are tied to their IDE's and it's wizzards and tools and not the lanugage itself, which they often have only a cursory understanding of how to use it.

[fs22]

I agree wholeheartedly with the previous posts. I am not a .Net(er), but I did look into it and weighed the pros and cons with CF and in my mind things like speed to press with CF (among other items) weighed heavily in CF's favor. I will continue to fight for CF using arguments as have been stated. If anyone does have some research/resources comparing switching to .Net and the costs involved, I will be happy to read through it.

Long live CF!

[tclaremont]

The most common argument I hear in favor of .net is related to cost. A very misguided argument at that.

People consider .net to be "free". Adobe ColdFusion Server starts at around $1200 US assuming you are not upgrading. (Notice I left out New Atlanta's Blue Dragon cost model)

What people don't understand, or don't WANT to understand is that development is an order of magnitude FASTER with CF, and with that being the case, the server license is paid for in what.... one week?

Good ColdFusion programmers, based on what I have seen, are more expensive per hour than net programmers. What this fails to acknowledge is that the increased speed of development results in a reduced number of billable hours. Therefore the total development cost of a project tends to be lower.

Pundits cannot seem to get past the fact that .net is "free" and Adobe ColdFusion server is not. "Free" is in quotes because even the very definition of "Free" is open for debate. Many of the add-ons for .net cost money. Similar functionality is "free" in ColdFusion.

Anyway, posting this question on a dedicated CF board is not going to get you the objective answers that you seek.

[TLC-IT]

Let's face it:  there are plenty of practitioners out there who have a fairly limited breadth of experience.  They know (say...) "dot-Net and absolutely nothing else (yet)."  And, instead of being willing to embrace new ideas, new tools and techniques, they wave their hands and mumble something about "market share," as though "follow the Lemmings" was an appropriate way to make strategic technical decisions.

Every company is different.  Every company's technical situation is different.  On top of this, tools are constantly evolving and growing more powerful.  The desire to exploit these new capabilities is tempered by the realization that there is no silver bullet.  When you develop a new piece of software for internal use, the maintenance and upkeep of that tool will be the predominant cost; not the cost of its initial development.  (Furthermore, you can bet your bank on the assumption that, whenever a developer (especially a younger one...) gets a better offer, it's "buh-bye!!" to the next greener pasture and you've got two weeks (if you are lucky) to manage to do without him or her.)  Amidst all of this, "you've got a business to run."  Software, and IT in general, is a cost to you.

So...  when presented with a new technology option, wise companies do a thorough cost/benefit analysis -- and the outcome of such an analysis strongly favors "status quo."  It just works out that way.  (Almost) every time.

I would also make this observation about ColdFusion:  "it is a shrewdly designed tool."  At first blush, you might not see what all the fuss is about.  Then you start to look more closely at its very novel approach of "just-in-time compiling" against a deep base-language (Java), and its very cleverly chosen set of tags and built-in abilities, and then you start to see a sparkling glint where you previously saw only dirt.  There are unique powers in this architecture that do not immediately jump out at you and demand, "look at me!"  Because they don't need to.  You will find them...

Dot-Net is powerful:  no one will deny this.  And I am sure that almost everyone here is thoroughly versed in it as well as ColdFusion.  We are very likely responsible for simultaneously supporting both technologies.  (And probably less-than-pleased, but resigned to it.)

[fs22]

Excellently stated...

[Guest]

I have recently been investigating serious jRun Memory leaks.  I'm finding very little effective knowledge on how to truly trace the code in order to identify and alter/repair the actual problem. 

I have an 801 server in production and updating the security patches and hot fixes, and (failing to) tune jrun and identify its memory leaks has been an incredibly frustrating process. 

I don't know if the .nuts are having any better luck maintaining and debugging their servers, but having worked with CF 5 thru 8 I can say that if 9 is not a departure from the releases of Macromedia's days then I may be looking at other server technologies soon. 

Just 9 months ago CF and jRun were neck and neck to be the second worst security ridden / attacked applications out there, only just behind another Adobe product.  Thats a pretty serious risk to weigh when considering what server technology to use.

http://www.communities.hp.com/securitysoftware/blogs/spilabs/archive/2009/08/31/top-five-web-applica...

I need a product that installs like the Filezilla client and updates as easily.  Why should it be so hard to update this software correctly?  Every one else is doing it : -)

I don't know how insecure .Net is, and I'm not here to just bash CF because... I prefer it, after all, to the MS technologies.

We have a large investment in CF 8 code and to watch it die on the server inside of jRun without being able to fix or even adequately diagnose it is taking all the fun out of doing what I love most!  I had to write a scheduled batch script to kill jRun if it reaches 950 MB and log an event to the Applicaiton Event Log because it was taking too many man hours to monitor and kill jrun during high production use.

Has anyone got any suggestions?  Will anyone at Adobe take my suggestions above? 

I appologize if this sand bags your quest to keep CF but these are real concerns with CF, for me at least ...

Finally, over anything I've said so far, I love CFML/CFScript.  The cost effectiveness of anyone coding in it generally beats other proprietary solutions at least from the productivity standpoint of an engineer (not withstanding engineers who are massively productive in any language) , but the licensing costs and the overhead of the issues I've been facing may force me in another direction.

Thanks for reading,

D.

[Adam Cameron.]

I have recently been investigating serious jRun Memory leaks.  I'm finding very little effective knowledge on how to truly trace the code in order to identify and alter/repair the actual problem. 

I have an 801 server in production and updating the security patches and hot fixes, and (failing to) tune jrun and identify its memory leaks has been an incredibly frustrating process. 

This is a bit of a thread hijack, but however.

Are you sure it's a memory leak in JRun, or is it your code just consuming too much memory?  This is a reasonable question, because usually when people claim it's a memory leak, it's actually not.

Have you run CF8's server monitor or FusionReactor over your site to have a look at what's using up RAM and when?  (Actually I'm only guessing CF's server monitor can do this... I've never used it... but FR certainly can).  Even running JConsole would probably be of some use sorting this sort of thing out.

What troubleshooting steps have you taken to sort this out (perhaps best to start a new thread for this, if you follow it up)?

Just 9 months ago CF and jRun were neck and neck to be the second worst security ridden / attacked applications out there, only just behind another Adobe product.  Thats a pretty serious risk to weigh when considering what server technology to use.

http://www.communities.hp.com/securitysoftware/blogs/spilabs/archive/2009/08/31/ top-five-web-applic...

Those injection vulnerabilities are hardly the fault of CF or Flex!  They're just hallmarks of bad programming practises, and anyone developing any app that takes user input and stores or redisplays it can fall afoul of that.  If one accepts user input, one needs to sanitise it before re-displaying it or re-using it.  That's common sense.

I am - by no means - a yes man for CF or other Adobe products (CF & CFB are the only Adobe product I use, actually... and I'll only be using CFB whilst my free trial works), but claiming those problems are intrinsic to CF or Flex is just dumb.

--

Adam

[tclaremont]

Having those kinds of issues with CF these days says a lot more about the programmer/administrator than it does about the platform.

[j_beckton]

Might I add that this is typical of what kind of dev you get in the procedural hack world of CF. I agree with you that it's the dev's responsibility to cover these types of holes in their code. In the .Net world your less likely to have wanna be software developers because they rarely make it past the learning curve of the platform. This is where CF comes in, it's very easy to learn and implement with little or no experience or understanding to good software design practices.

[Owainnorth]

Oh man, I'd have gotten up earlier had I realised this old chestnut was kicking off again.

Ever heard the phrase "horses for courses"? It's a stupid one I know, but it applies here. ColdFusion and .NET are two different things. In fact, .NET isn't even a language it's a framework, so talking about it so simply isn't even a fair comparison. In order to create a basic HTML page in CF you need to know HTML and CF. To use a .NET page, you need to know HTML, ASP.NET and another language for the backend - normally C# or VB. So that's an extra bit of learning and development. Show someone how to knock up a quick SQL query and CF page? Fair. Do the same with C#? Start explaining about classes and typing? Compilation? Namespaces? Good luck with that.

Personally, I use both. I use C# whenever I want a strongly-typed application that integrates well with WIndows, Active Directory, fileshares, and can run as a Windows Service or exe. But you simply cannot knock how fast CF is to develop. Where I work, we have an internal system that the staff use for reporting, processing orders and such - this is written entirely by myself. Someone can ask me for a new page, and within five minutes I can have the CFCs edited, page created, menu XML updated, code up to a datacentre and replicated across our loadbalanced farm.

CF Development time pisses all over .NETs I'm afraid. And I'm not saying that as a CF fanboi, I actually prefer using the .NET platform and writing C#. As long as I have the time.

As for costs, that's a skewed view as well. CF can run perfectly on Linux, negating the need for expensive Windows licenses.Buy the Enterprise version and install ESXi on a server, you can have as many Linux CF boxes as you want for no extra cost.

Horses for courses. Let's all move on with our lives.

[j_beckton]

Unless your ASP.Net apps are running on 2.0 and your stuck with VS 2003 or 2005 I can see why you would state that dev time for CF pisses all over .Net. However with the current release of the .Net framework as well as Visual Studio that is no longer the case. Unless you do not know how to use the tooling.

I can build a web page that hits a database just as fast in VS 2010 as I can with any CF IDE, and that includes deployment as well. With both platforms it takes a bit of preparation but none the less CF pissing on .Net dev time is a thing of the past.

While on the subject of dev time, is faster really better? Any one can create a cfm page, slap a cfquery tag inside with some basic sql then a cfoutput and call it an app but is that really the right way to build software? If your not using standard design patterns in your CF apps to separate concerns then hell yeah you can build an app in CF twice a fast. But then what happens when you have to refactor a bunch of it to add new features? Your dev time will nearly double.

I typically use an MVC pattern for my CF apps and my model contains a data access layer and a service layer. My service layer uses a DTO or Bean to communicate data across layers. Using a layered pattern in CF is no less time consuming to build than in C#, you can try to argue that but if you do your either fooling yourself or your not using OOP patterns correctly.

As far as maintenance .Net & Visual Studio pisses all over any CF platform tool. We all know that maintenance is where most of the investment is for most any software project. While I used to pull my hair out trying to refactor other peoples procedural or mock OOP CF messes I can breeze through the same tasks with VS on the .Net platform in half the time.

I am not sure where CF being easier to learn makes it better... unless your programmers have little or no programming experience.

And as far as your new page scenario, the same is possible in ASP.Net as well with the same amount of preparation. Personally I would never make changes to a production app and push them up to the live server all within 5 minutes. I typically like for my code changes to go through atleast a simple QA process performed by another individual before they go live all of which is typically impossible within 5 minutes. This may not be needed if your working on really simple apps but in my case I work with fairly complex software requirements and stake my reputation on code that works.

[Owainnorth]

with the current release of the .Net framework as well as Visual Studio that is no longer the case. Unless you do not know how to use the tooling.

Exactamundo. I've been using VS for a couple of years now, wouldn't have a clue how to do any of that. I have, however, mastered notepad

While on the subject of dev time, is faster really better?

Not always. But sometimes, yes.

I typically use an MVC pattern for my CF apps and my model contains a data access layer and a service layer. My service layer uses a DTO or Bean to communicate data across layers. Using a layered pattern in CF is no less time consuming to build than in C#, you can try to argue that but if you do your either fooling yourself or your not using OOP patterns correctly.

But what if you don't want to use MVC, OOP and all that guff? What if you just want a page, with a SQL query and a report?

I am not sure where CF being easier to learn makes it better... unless your programmers have little or no programming experience.

Who the hell are "my" programmers? All programmers start with no programming experience - that's kinda my point. And from that point in time, I guarantee you 100% of people could learn how to do a basic CF page quicker than an ASP/C# one.

Personally I would never make changes to a production app and push them up to the live server all within 5 minutes. I typically like for my code changes to go through atleast a simple QA process performed by another individual before they go live all of which is typically impossible within 5 minutes. This may not be needed if your working on really simple apps but in my case I work with fairly complex software requirements and stake my reputation on code that works.

Oh I see, you're one of *those* programmers

As I said in my last post, I'm genuinely not on either "side", because I don't consider them to be competitors - I use both because they're very different products. And because sometimes all you want is a simple page, with a simple query and a quick dump of the data, And CF simply does that better. Sometimes you want to throw some acronyms at it and go MVC, OOP, WGAS all over it, in which case .NET is far better, as it's strongly typed.

Anyway, boring.

[j_beckton]

Oh well if you just want to build "Web Pages" ASP.Net has that covered too... Brand new stuff and just as fast and easy as ColdFusion in regards to developing Web Pages. Plus you still have the power and features of the .Net framework at your disposal.

http://www.asp.net/web-pages

[ilssac]

Then please show me a .NET equilavent of this code that does not take A LOT more lines of code.  Because I sure couldn't find any examples with internet searches when I needed one a couple of weeks ago.

<cfhttp url="https://ets.cdpr.ca.gov/pur-loader-dev/ws/XXXXX.cfc" method="post" multipart="yes">
    <cfhttpparam name="authentication" value="XXXXXXXXXXX" type="formfield">
    <cfhttpparam name="ppurfile" file="#expandPath('aFile.xml')#" type="file">
    <cfhttpparam name="method" value="Submit" type="url">
</cfhttp>

[j_beckton]

no problem.. Actually dealing with web services is where the ASP.Net tooling really shines. In CF once you get your response back from that service you'll have to deal with parsing the xml into a data type that you can work with like a struct or something or else you can let CF parse it into a struct if you do not mind the complicated struct it will give you to work with.

In .Net the tooling generates the proxies and everything for you based on the WSDL of the service so that the service request and response is automatically parsed and presented in a correct data type, going out xml coming in a collection.

here is an example of consuming a web service, you will have to scroll as the first part shows you how to create the service. Just look for "Scenario 2: Consuming a web service from an external source."

http://ctrlf5.net/?p=66

When your done letting Visual Studio create the needed proxie code allowing your app to talk to the service all you have to do to use it is...

var myService = new MyNameSpace.MyServiceClient();
var myParmsObj = new MyNamespace.MyParmsObj();
myParmsObj.Authentication = "xxxxxxxx";
myParmsObj.Ppurfile = "aFile.xml";
myParmsObj.Method = "Submit";

MyNamespace.MyServiceResult myServiceResult = myService.RemoteCFCMethod(myParmsObj);

[Dave Watts]

In your web service example, note that it's about consuming ASP.NET web services from within ASP.NET. If you consume CF web services from CF, native objects are preserved just fine. It's when you cross boundaries that aren't defined by the SOAP specification and XML Schema datatypes that you run into the problems you describe. But even then, you can drop down to Axis directly to deal with those problems in most cases.

Dave Watts, CTO, Fig Leaf Software

http://www.figleaf.com/

http://training.figleaf.com/

[Dave Watts]

Oh, also, you're assuming that the CFC is being exposed as a SOAP service. In ilssac's example, it isn't being used as a SOAP service. Not that you can't invoke REST services from .NET, just thought I'd point that out.

Dave Watts, CTO, Fig Leaf Software

http://www.figleaf.com/

http://training.figleaf.com/

[ilssac]

j.beckton wrote:

no problem.. Actually dealing with web services is where the ASP.Net tooling really shines.

That may be nice, but we are not dealing with an actual web service here.  My example was a HTTP FORM MULTI-PART FILE POST.  Thus the <cfhttp...> tag with the multipart parameter and the <cfhttpparam...> tags with the type="formfield", type="file", and type="url" parameters.  As best as I can tell, your example does not upload a file, it just provides a string of the file name.

That is what I could not find an example.  I could find a simple example if one ONLY wanted to post the file.  Doing a non-file form post was not too bad.  But combining file and non-file fields in the http post turned out to be rather difficult and too time consuming for the effort the task at hand warrented.

[Dave Watts]

"Unless you do not know how to use the tooling."

Well, learning how to use the tooling is a significant cost to ASP.NET development. The cost of "learning the tooling" is part of the cost of development. I do development in J2EE, .NET, and CF (in roughly that order nowadays), so I think I have a fair idea of that cost. I think that ordering also reflects the overall complexity of that development: I find J2EE the most complex, followed by .NET, followed by CF.

And that tooling involves some implementation abstractions which, arguably, hurt the developer when those abstractions need to be pulled away. For being a high-level language, CF doesn't rely all that heavily on these implementation abstractions. For example, in .NET I've occasionally had to dig under things like VIEWSTATE for reasons I won't bother going into here.

Dave Watts, CTO, Fig Leaf Software

http://www.figleaf.com/

http://training.figleaf.com/

[j_beckton]

Well said Dave but typically unless your looking at switching platforms an employer does not look at what the developer needs learn and how much it will cost in order to use that platform. I know that the original post mentioned they are considering .Net over CF so factoring in retooling your devs would apply but in a different context, when you hire developers typically you hire the ones that have experience in the platform your using or going to use. The cost of retooling existing developers is not always a factor and should not be considered when comparing two platforms.

As a developer if you want more skills in other platforms you have to deal with that expense on your own and not expect an employer to do it.

Luckily MS now has an MVC framework so cludgy abstractions do not have to be a problem anymore. I tried WebForms and did not like it, ASP.Net MVC has been a great pleasure to work with coming from CF.

[Dave Watts]

I had no idea you could speak for all employers, or could definitively state that the cost of retooling developers shouldn't be considered when comparing platforms. If only life were so simple.

Very often, developers not only have to know their tools, but also the problem domain for which they're developing. That knowledge can be very valuable to employers, and more difficult to learn than the tools themselves. So, no, you don't get to state which costs are worth considering, and which aren't. That's just not how the world works.

Dave Watts, CTO, Fig Leaf Software

http://www.figleaf.com/

http://training.figleaf.com/