Watch What You Write, Someone Is Reading

Today I received the following comment here, on an older post on Variables and Naming Conventions:

...I wish Adobe would publish and adopt some kind of official naming convention. Sometimes reading sample code written in some other convention can make things harder to follow...
It was almost funny that this comment had come in when it had. Recently I was doing a lot of research for a User Group presentation I just did on the new ColdFusion 8 Ajax Components (have to re-record it before public release). In the process, I spent a great deal of time going over documentation all over the internet, from LiveDocs to countless blogs, absorbing the wealth of information that is already out there. It was outstanding that there were so many resources out there for people to learn from. On the other hand, it was a little sad that so much of the sample code was written in ways that can really start new developers off with some bad habits.

I'm not perfect, by any means, but I try to pay careful attention to the code that I place on this blog for readers to use and learn from. One thing that I attempt to do is pay attention to basic Web Standards, like using XHTML (the current standard) instead of HTML, keeping styles in the stylesheet, and having unobtrusive JavaScript. I don't always do it, sometimes it doesn't make sense for a quick example, but I try, especially within code downloads. I also try to adhere to my own Coding Guidelines, so that code appears to be consistent and easy to read and understand.

Probably the one that bothers me the most, and that I see most prevalent in blogs, documentation, and books, is the lack of proper variable scoping. I know that, often, we're just publishing quick examples, but this can be an extremely detrimental practice. I have worked on some very large enterprise applications, with years of code written by half-a-dozen different developers, most of whom learned their ColdFusion (and development) skills through the docs or a book. Many had actually come up with some very creative and effective algorithms to fix some issue, or create some new whiz bang feature, but their code was so poorly scoped that, after time, it could take down the server. Why? How? Enterprise sites may contain several hundred (or thousand) templates, containing dozens of variables on each page, and can potentially be hit by hundreds (or thousands) of users simultaneously. Multiply the number of variables by the number of pages by the number of users, then imagine ColdFusion doing a ScopeCheck on each one, to figure out which scope each variable requested belongs in. Even if the variable is in the VARIABLES scope, it's still that many times ScopeCheck will be called while rendering a page.

Still not convinced? Go download varScoper, and run it on your project root folder, including your subfolders, and see what it comes up with. Yeah, I'm still in shock. Cleanup on that is easier on a small subproject scale, but it's definitely forced me to think better when I'm writing my code, paying attention as I go, to minimize the performance impact of my applications, no matter how small it may be. I learned my bad habits from the docs, various books, sample code slung around on the CF-Talk list. I've continued to realize that there are better ways of doing things (like OOP and frameworks), and adjust my style and methods, and I think it's important to consider these 'best practices' when contributing. A little more code, but the right thing to do in the end, for you, your app, and your systems.

So, if you own a site of documentation, revise it. If you're writing a book, edit it. If you publish a CF blog, live it. The up-and-coming are reading us all of the time to find out how to use this wonderful language. Let's try to show 'em how to do it the right way. You might not follow any guidelines at all, within your development, but this scoping thing is way too important to gloss over, and will only help everyone in the long run.

Yes, I Am King!

Well, this was fun! Thanks Aaron.

NerdTests.com says I'm an Uber Cool Nerd King.  What are you?  Click here!

CFGrid Gotcha

So, I'm finally playing with some of the new Ajax controls built into ColdFusion 8. They're based on ExtJS (for the most part), and I thought it would be cool to dig in and see what I could do.

So, pulled up the documentation. First I built a basic CFC, with a remote access method that pulls all of the records from the Art table of the cfartgallery db. Then I built the display page, with the cfgrid and cfgridcolumn tags. I used the bind attribute to bind the grid to the cfc method. Tried it out and...error.

view plain print about
1CFGRID: Response is empty [Enable debugging by adding 'cfdebug' to your URL parameters to see more information.]

OK. Fun. No response messages showing in Firebug, but the right parameters were getting passed through. Google is your friend, right? One reference that I could find, in the comments on a post at Ben's site, but it only pointed me towards the Application.cfc, with no explanation on what the problem was or how to fix it.

So, I changed the file name of my Application.cfc. I didn't need it this early in the game, so I took it out of play. Voila! It works. OK, so what's in the Application.cfc?

Well, I had already commented out the onError method (figuring out an issue with Coldspring). There wasn't any output in any of the methods. I went over all of my attributes and mappings...nothing. Then I noticed something.

I took the Application.cfc template from Ray's site, with very minor adjustments. I finally noticed that one function didn't have an 'output' attribute, onRequest.

view plain print about
1<cffunction name="onRequest" returnType="void">
2        <cfargument name="thePage" type="string" required="true" />
3        <cfinclude template="#arguments.thePage#" />
4    </cffunction>

Once I commented this function out the call worked perfectly. Well, lessons learned...

ColdFusion 8 Fun: Looping Files

OK, so I've been working on my mother's website for...well, too long. One of the reasons is I've been waiting on her to get approval to get a feed of listings, so we can put them directly on her site. Well, she finally got the approval, so I've been having fun this weekend, pulling in data and images, setting up database tables. The Works.

These feeds are tab delimited text files. The first line being a listing of all of the columns, with all of the rest being the data. So, I set up a staging table, with column names that match those in the file (luckily they provide a listing of the columns, along with their data type and length, in a separate .log file). Next, I used the Illudium PU-36 Code Generator to quickly give me some data access objects, and then settled down to write a little code.

Now, my first file has 7,000+ records in it, so I go ahead and give myself a little time for the code to do it's job.

view plain print about
1<cfsetting enablecfoutputonly="true" requesttimeout="600" />

Next thing I wanted were a few variables and objects to work with.

view plain print about
1<cfset VARIABLES.lineNum = 1 />
2<cfset VARIABLES.filePath = expandPath(".") & "\myFile.txt" />
3<cfset VARIABLES.Bean = CreateObject("component","feedRecord") />
4<cfset VARIABLES.DAO = CreateObject("component","feedRecordDAO").init(APPLICATION.dsn) />

And then I setup the loop on the file.

view plain print about
1<cfloop file="#VARIABLES.filePath#" index="VARIABLES.line">
2    <!--- Code to go here --->
3</cfloop>

OK, for those who don't know, the DAO object that is created by the code generator takes a bean object as the argument for the save() method. The bean object has an init() method with all of the column names as non-required arguments. So, how to best initialize my bean? Well, the data file's first row is a tab delimited list of the column names, so I decide to use it. First, I only want the first row to give me a data structure of the column names, in the order I'll need them. Hmmm? Ok, I decide to use an Array.

view plain print about
1<cfloop file="#VARIABLES.filePath#" index="VARIABLES.line">
2    <cfif VARIABLES.lineNum gt 1>
3        <!--- This is for later --->
4    <cfelse>
5        <cfset VARIABLES.propOrder = ArrayNew(1) />
6        <cfset VARIABLES.lineCount = 1 />
7        <cfloop list="#VARIABLES.line#" index="VARIABLES.listItem" delimiters="#Chr(9)#">
8            <cfset VARIABLES.propOrder[VARIABLES.lineCount] = VARIABLES.listItem />
9            <cfset VARIABLES.lineCount++ />
10        </cfloop>
11        <cfset VARIABLES.lineCount = 0 />
12    </cfif>
13    <cfset VARIABLES.lineNum++ />
14</cfloop>

Notice that the first part of my flow control is currently blank. This area I left at the beginning, as most lines will meet this criteria, and that's where the meat of the processing will be handled in the end. This Array, though very important, is only handled on the first row of the file. It will process first, because of the way the flow control is written, but bypassed throughout the rest of the process. BTW, I love the JS style operators;)

Now, I used an Array to maintain the order of the key names, but ultimately I'll need a Struct to pass into the bean's init() method, as an argumentCollection.

view plain print about
1<cfloop file="#VARIABLES.filePath#" index="VARIABLES.line">
2    <cfif VARIABLES.lineNum gt 1>
3        <cfset VARIABLES.resProp = StructNew() />
4    ....

Now, I was going to list loop through each line to set my Struct, but found out the hard way that <cfloop> still doesn't like empty items in a string. I was getting errors all over the place about truncated data and what, before I noticed data wasn't in the right place. What to do? Take a different approach! Instead of looping a list, I'll loop an Array, and make my Array from the list, while using the new includeEmptyFields option.

view plain print about
1<cfloop file="#VARIABLES.filePath#" index="VARIABLES.line">
2    <cfif VARIABLES.lineNum gt 1>
3        <cfset VARIABLES.resProp = StructNew() />
4        <cfset VARIABLES.arrProps = ListToArray(VARIABLES.line,Chr(9),true) />
5        <cfloop from="1" to="#ArrayLen(VARIABLES.propOrder)#" index="VARIABLES.itemCount">
6            <cfset VARIABLES.resProp[VARIABLES.propOrder[VARIABLES.itemCount]] = VARIABLES.arrProps[VARIABLES.itemCount] />
7        </cfloop>
8</code>
9
10Did you see it? Simple, eh? Now I have a Struct, where the data from each line matches up with the keys set from the first line of the file. All that's left is to set my bean and pass it to the save() method of the DAO.
11
12<cfloop file="#VARIABLES.filePath#" index="VARIABLES.line">
13    <cfif VARIABLES.lineNum gt 1>
14        <cfset VARIABLES.resProp = StructNew() />
15        <cfset VARIABLES.arrProps = ListToArray(VARIABLES.line,Chr(9),true) />
16        <cfloop from="1" to="#ArrayLen(VARIABLES.propOrder)#" index="VARIABLES.itemCount">
17            <cfset VARIABLES.resProp[VARIABLES.propOrder[VARIABLES.itemCount]] = VARIABLES.arrProps[VARIABLES.itemCount] />
18        </cfloop>
19        <cfoutput>Saving Record ## #VARIABLES.lineNum#. </cfoutput>
20        <cfset VARIABLES.Bean.init(argumentCollection:VARIABLES.resProp) />
21        <cftry>
22            <cfif VARIABLES.DAO.save(VARIABLES.Bean)>
23                <cfoutput>Record saved.<br /></cfoutput>
24            <cfelse>
25                <cfoutput>Error saving record.<br /></cfoutput>
26                <!--- custom cfthrow here --->
27            </cfif>
28            <cfcatch type="any">
29                <!--- and a custom error handler here --->
30            </cfcatch>
31        </cftry>
32        <cfflush />
33    <cfelse>
34        <cfset VARIABLES.propOrder = ArrayNew(1) />
35        <cfset VARIABLES.lineCount = 1 />
36        <cfloop list="#VARIABLES.line#" index="VARIABLES.listItem" delimiters="#Chr(9)#">
37            <cfset VARIABLES.propOrder[VARIABLES.lineCount] = VARIABLES.listItem />
38            <cfset VARIABLES.lineCount++ />
39        </cfloop>
40        <cfset VARIABLES.lineCount = 0 />
41    </cfif>
42    <cfset VARIABLES.lineNum++ />
43</cfloop>
44<cfsetting enablecfoutputonly="false" />

That's it! Nothing to it! Now, there are probably better ways, and half of this should be encapsulated even further, and it will break if the feed provider changes the column names. But, hey, it was fun! Right?

Example code included below with the Download link.

Up To The Latest And Greatest

OK, last night I upgraded to BlogCFC 5.9, which Ray released the other day. Pretty heavy update, with over 30 file changes in the readme, but very smooth when using WinMerge (thanks to Mark Drew for that little tip).

One item of interest, though. Ray mentions in the readme that he's stopped logging the changes in the document headers, stating that it's redundant to place them there and within the readme doc. He also states "I decided I'd skip that since BlogCFC 6 will have new files." I also noticed, while reviewing the changes, that he has code in place for checking the server version. CF 8 specific changes on the way? We'll have to wait and see.