Sometimes it is necessary to write different code for Adobe ColdFusion vs Railo vs OpenBD, due to differences in implementation.
Do people have a specific method that they use for this?
For example, switching on ProductName is one option:
<cfswitch expression="#Server.ColdFusion.ProductName#">
<cfcase value="ColdFusion Server">
...
</cfcase>
<cfcase value="Railo">
...
</cfcase>
<cfcase value="BlueDragon">
...
</cfcase>
</cfswitch>
Is that the best way, or can anyone suggest something better?
When you get down to it, that's probably the most reliable way. You might be safer doing feature detection rather than explicit product checks, but CFML doesn't have a lot of introspection features you can use for that kind of thing.
If you use CFCs in your work, then you can also hide some of these alternate implementations using patterns like Strategy and Template Method. But when you're choosing an implementation, you'll probably still come back to ProductName.
i think a better way to handle this would be to create a cfc for each engine and a matching method name in each cfc. then what you could do is invoke that cfc and run the method depending on the engine. you can use your switch statement in the onapplicationstart() event to set an application wide variable to initialize and store the engine specific cfc. a small example would be:
<cfset loc.engine = "adobe">
<cfswitch expression="#Server.ColdFusion.ProductName#">
<cfcase value="Railo">
<cfset loc.engine = "railo">
</cfcase>
<cfcase value="BlueDragon">
<cfset loc.engine = "openbd">
</cfcase>
</cfswitch>
<cfset application.engine = createobject("component", "engines.#loc.engine#").init()>
then in your code all you would have to do:
<cfset myvar = application.engine.somemethod(arguments)>
granted it's still not the prettiest solution, but at least you will be containing all the engine specific code in one place and not littering your codebase with switch logic.
Related
I wrote the following function:
<cffunction name="check_session_valid" returntype="boolean">
<cfif NOT StructKeyExists(session,"username") OR (len(session.username) EQ 0)>
<script>location.href = 'logout.cfm'</script>
<cfabort>
</cfif>
<cfset session.myApp_start = now()>
<cfreturn true>
</cffunction>
In my .cfm page, I can call that function using
<cfset session_valid = application.lib.check_session_valid()>
OR
#application.lib.check_session_valid()#
What's the difference? Best practice?
Since you asked about best practice, which is a matter of opinion, I think you can improve your function by having it returning either true or false depending on whether or not session.username exists and has a length greater than 0. Then you can use it like this:
<cfif application.lib.check_session_valid()>
code for this condition
<cfelse>
<cflocation href = "logout.cfm">
<!--- note that cfabort is not necessary --->
<cfif>
Regarding your specific question, I think the extra variable, session_valid, is a waste of typing. However, that is simply my opinion.
Not related to your question, I found it curious that you would direct users to a page called logout.cfm. Often users are directed to a page that allows them to log in.
To be honest, both are valid and both would be considered best practice depending on what you are trying to do.
My rule of thumb is if I will need to use the result of a function call more than once, I will set it to a variable
myResult = application.lib.check_session_valid();
If I will only need to use the variable once I would do what Dan mentioned
if( application.lib.check_session_valid() ){
// Do stuff
}
The difference between the examples you showed are
<cfset session_valid = application.lib.check_session_valid()>
This will set the variable named session_valid to whatever is returned from the call to check_session_valid().
#application.lib.check_session_valid()#
This will, in .cfm pages, simply render the value returned from the call to check_session_valid() assuming it is inside of a <cfoutput> tag. There are other places this would also render the value, such as inside a <cfsavecontent>.
I was curious if there is a way to force a ColdFusion tag to hold an attribute as default, such as the datasource in cfquery.
For example instead of writing
<cfquery datasource="mydatasource">
I can write
<cfquery>
and the system automatically knows that the datasource is "mydatasource".
Would be really cool if this was possible.
It is actually possible for datasource, but not for everything.
You may set a this.datasource="mydatasource" as the default datasource in your Application.cfc
https://wikidocs.adobe.com/wiki/display/coldfusionen/Application+variables
The practical answer to your question are the custom tags. You can extend the features of ColdFusion tags to match your needs.
Taking into example the cfquery tag and wrapping a custom tag around it. Provide all the default values you want for the parameters of the cfquery into the tag's attribute default.
So essentially your custom tag page would be something like:
flexiquery.cfm
<cfif THISTAG.ExecutionMode EQ 'end'>
<cfparam name="Attributes.datasource" default="someDSN">
<cfparam name="Attributes.cacheWithin" default="#CreateTimeSpan(0,6,0,0)#">
<cfparam name="Attributes.maxRows" default="25">
<cfparam name="Attributes.timeOut" default="600">
<!--- some logic you want to perform --->
<cfquery datasource="#Attributes.datasource#"
cacheWithin="#Attributes.cacheWithin#"
maxRow="#Attributes.maxRows#"
timeOut="#Attributes.timeOut#"
<cfoutput>#THISTAG.GeneratedContent#</cfoutput>
</cfquery>
<!--- Caller assignment and other processing --->
</cfif>
And now you can use it and re-use it across your project, the way you wanted and even overriding the value you want to be different:
<cf_flexiquery>
<!--- you query here --->
</cf_flexiquery>
or
<cf_flexiquery maxRows="100" timeOut="1200">
<!--- you query here --->
</cf_flexiquery>
It gives you a fair idea of how to go with it. I have extended the custom tags features to leverage the features of cfhttp, cfpdf, cffile etc.
This is only way you can adopt the flexibility you want with ColdFusion tags and it works perfectly.
I am using ColdFusion 9.0.1
I am taking over a site and the guy before me created about 100 variables and put them into the APPLICATION scope. I believe that his 100 variables were continuously being overwritten with each page load.
Basically, he had this in Application.cfc:
APPLICTION.VariableOne = "SomeStringOne";
APPLICTION.VariableTwo = "SomeStringTwo";
APPLICTION.VariableThree = "SomeStringThree";
My plan is to keep it simple and yet very readable is to test for a specific structure in the application scope. If it's not there, create the structure and variables:
if (not isDefined("APPLICTION.AppInfo") or not isStruct(APPLICTION.AppInfo)) {
APPLICTION.AppInfo = structNew();
APPLICTION.AppInfo.VariableOne = "SomeStringOne";
APPLICTION.AppInfo.VariableTwo = "SomeStringTwo";
APPLICTION.AppInfo.VariableThree = "SomeStringThree";
}
Of course, once the site is live and we are done creating all of the application variables, I'd move this into the into the onApplicationStart() method.
The solution that I want has to be more about "readability" and less about "efficiency". Several non-CFers, but very experience coders will be using this and will need to "get it" quickly.
Does my plan have any gaping holes or is it too inefficient?
Is there a more readable way of creating and managing application variables?
Why not move the definition into onApplicationStart() right now? If you need to reset them during development, you could always pass in a URL variable to flag it for reset, like so:
<!--- in Application.cfc --->
<cffunction name="onRequestStart">
<cfif IsDefined("url.resetApp")>
<cfset ApplicationStop()>
<cfabort><!--- or, if you like, <cflocation url="index.cfm"> --->
</cfif>
</cffunction>
Actually, after re-reading the OP, and reading the suggested solutions, I'm going to have to agree with the OP on his setup, for this very important reason:
This, in onApplicationStart()
APPLICTION.AppInfo = structNew();
APPLICTION.AppInfo.VariableOne = "SomeStringOne";
APPLICTION.AppInfo.VariableTwo = "SomeStringTwo";
Can then later be turned into this, within onRequestStart()
<cflock name="tmp" type="readonly" timeout="15">
<cfset REQUEST.AppInfo = APPLICATION.AppInfo />
</cflock>
Your app can then go on to access the REQUEST vars conveniently, esp, if you decide you want to cache CFCs in the same scope--they would simply go into a separate key:
APPLICATION.Model.MyObject = CreateObject('component','myobject');
Which, of course, also gets poured into REQUEST (if you choose)
Want to go Jake Feasel's route above? No problem:
<cfif isDefined('URL.reload')>
<cfset APPLICATION.Model = StructNew() />
</cfif>
Now you're able to flexibly kill your object cache but maintain your vars (or vice versa as you choose).
This is a great setup for another reason: If you want to build in your own Development/Production "mode", in which the development mode always recompiles the CFCs, but the production mode keeps them cached. The only change you have to make on top of this, is the REQUEST set noted above:
<cfif (isProduction)>
<cflock name="tmp" type="readonly" timeout="15">
<cfset REQUEST.AppInfo = APPLICATION.AppInfo />
</cflock>
<cfelse>
<cfset REQUEST.AppInfo = StructNew() />
<cfset REQUEST.AppInfo.VariableOne = "SomeStringOne" />
...etc...
</cfif>
You can also make the setting of vars and the creation of objects into a private method within Application.cfc, for even further convenience.
I would go ahead and just use OnApplicationStart but back in the pre Application.cfc days we used to do something like Application.Build and if the Build value was different then we did all of our sets on Application variables. So quick and dirty would be something like:
<cfparam name="Application.Build" default="" />
<cfset Build = "28-Nov-2011" />
<cfif Application.Build IS NOT Variables.Build OR StructKeyExists(URL, "Rebuild")>
<cfset Application.Build = Variables.Build />
<!--- A bunch of other CFSETs --->
</cfif>
This method though was something we used back when all we had was the Application.cfm
Let's say i've just parsed someone else's XML document which is a response to an API request. I want to know if a value nested deep inside exists. If my API request worked, it will be in the same place every time. If my API request fails, the root of the XML is very different.
If I try <cfif structKeyExists(myStruct.level1.level2.level3, 'myTarget')> on a failed api request, I get the fatal error: Element LEVEL1.LEVEL2 is undefined in MYSTRUCT.
Of course, I could try to depend on the root level of the XML telling me of success or failure, and not looking for the result if it failed, but... barring that solution, what should i do?
Do i need to check for the existence of each level of the struct? As in:
<cfif structKeyExists(myStruct, 'level1')
and structKeyExists(myStruct.level1, 'level2')
and structKeyExists(myStruct.level1.level2, 'level3')
and structKeyExists(myStruct.level1.level2.level3, 'myTarget')>
<!--- ... --->
</cfif>
This is not a real-world problem, this is just something i've faced too many times. Please don't tell me solutions that involve changing the API or solutions like those in the third paragraph.
Thanks!
edit: i should have mentioned why i can't use isDefined() - some of the keys do not have syntactically valid names, so isDefined() throws an error, eg myStruct.level1[42].level3
XMLSearch
I would use the parsed XML document (i.e. xmlDoc) and XMLSearch:
<cfset xmlDoc = xmlParse(responseData)>
<cfset nodes = XmlSearch(xmlDoc, '/level1/level2/level3/myTarget')>
<cfif arrayLen(nodes)>
<!--- do something, you have the "nodes" array to work with too --->
</cfif>
xpath for XMLSearch() assumes the structure keys are nodes. You would need to modify accordingly if, for instance, 'myTarget' is an attribute of a node.
StructFindKey
Another way of doing this would be StructFindKey.
<cfset result = structFindKey(myStruct, "myTarget")>
<cfif arrayLen(result) AND result.path EQ "level1.level2.level3">
<!--- do something --->
</cfif>
Conclusion
Haven't tested, but I believe either will be faster than using IsDefined() or a try-catch block. Has the advantage over XMLValidate() of not needing a DTD. And, even with a DTD, the node you want may be defined as optional, so it could still validate.
You could validate the XML against a DTD to make sure the document was in the right format. XmlParse() and XmlValidate() both take a DTD as a parameter.
<cfset validateResult = XmlValidate(myXmlDocument, myDTD)>
<cfif validateResult.status>
<!--- xml is valid continue processing --->
<cfelse>
<!--- xml did not validate handle the error --->
</cfif>
Personally I wouldn't go crazy checking for every level of a 'deep' structure like this. I would presume that if the top level exists the rest of the document will be as you expect, and I'd just address the document from there.
If you wanted you could perhaps try to address the value in your struct and wrap it in a try/catch. That way you can handle any errors at any 'level' in the same way.
<cftry>
<cfset myVar = myStruct.level1.level2.level3 />
<cfcatch type="any">
<!--- Handle error --->
</cfcatch>
</cftry>
Hope that helps some.
I know I'm going to get booed off the stage here, but this is where isDefined() can save you a lot of typing:
<cfif isDefined(structKeyExists(myStruct.level1.level2.level3)>
<!--- do something --->
</cfif>
I know this is a year old, but I'm going to put in an answer here. I struggled for a good long time with this one, till I found a simple solution. If I know the structure of the XML already, a simple IsDefined works to test if the node or node attribute exists. I don't think most people know you can do this, or have tried and failed because they didn't include single quotes in the IsDefined function.
So say I grab some user xml from a web service somewhere and want to display the user's ID.
<cfhttp url="https://mycompany.com/mywebservices/getusers" username="me" password="mysecret">
<cfset userXMLDoc = XMLParse(ToString(cfhttp.FileContent).trim())>
<cfif IsDefined('userXMLDoc.Data.Record.User.XmlAttributes.id')>
<cfdump var="#userXMLDoc.Data.Record.User.XmlAttributes.id#">
<cfelse>
<cfoutput>Failed: No User ID found</cfoutput>
</cfif>
By default, ColdFusion passes simple types (like numeric, string, and GUID) by value to functions. I'd like to pass a simple type by reference.
I'm currently wrapping a simple value in a struct (they get passed by reference). This solves my problem but it is very ugly:
<!--- TheFunctionName---->
<cffunction name="TheFunctionName">
<cfargument name="OutVariable" type="struct">
<cfset OutVariable.ID = 5>
</cffunction>
<cfset OutVariable=StructNew()>
<cfset TheFunctionName(OutVariable)>
<!--- I want this to output 5--->
<cfoutput>#OutVariable.ID#</cfoutput>
I'd rather something like this:
<!--- TheFunctionName---->
<cffunction name="TheFunctionName">
<cfargument name="OutVariable" passbyref="true">
<cfset OutVariable = 5>
</cffunction>
<cfset TheFunctionName(OutVariable)>
<!--- I want this to output 5--->
<cfoutput>#OutVariable#</cfoutput>
AFAIK, there's no way to pass simple values by reference in ColdFusion. The only workaround I can think of is the one you're already using.
Instead, I would suggest trying to restructure your program to work with the grain of the language. In cases where there's only one simple value to "modify", you could just make your function return the new value, and call it like:
<cfset SomeVar = TheFunctionName(SomeVar)>
In cases where you're modifying multiple values, take a step back and think about whether it's possible to bundle those multiple values up into a CFC with your mutator functions becoming methods of the CFC. This could be clearer and more maintainable solution anyway.
You can arrange for the variables used outside and inside the function to be in a scope that exists in both code areas. For example, if you put a variable in the "session" or the "request" scope you will be able to access it from within the function. The changes made will persist.
Note that when you are doing this you aren't actually "passing" the variables to the function. The function just assumes the variable exists or creates it, depending on how you code it.
<cffunction name="TheFunctionName">
<cfset Request.StrVar = "inside function<br />" />
</cffunction>
<cfscript>
Request.StrVar = "outside function<br />";
WriteOutput(Request.StrVar);
TheFunctionName();
WriteOutput(Request.StrVar);
</cfscript>
About ColdFusion Scopes
If there is any doubt about the calling page declaring the variable in advance when it is required you'll have to do some legwork with the <cfparam> tag or IsDefined() function.
If you:
declare the function inside of a CFC
invoke the function using <cfinvoke>
You would be able to specify the <cfinvoke> parameter "returnvariable", and then output that variable however you like.
<cfinvoke component="this" method="TheFunctionName" returnvariable="blah">
<cfinvokeargument name="data" value="whatever" type="string">
<cfreturn data>
</cfinvoke>
<cfdump var="#blah#">
If you are writing everything in cfscript, then I would go with what SurroundedByFish said.