Amazing

I'm getting an error when trying to install the wheels-cli

Let me look into it, will let you know as soon as it is working

playing the jeopardy theme song in my head while I wait for the GitHub actions to finish testing and pushing the change

@David Belanger you should be able to do the install now.

Not directly related to Wheels, but my blog post on destructuring a query into nested objects was directly inspired by the fact that CFWheels does this when you

include / returnAs=structs

in your queries - which is a very cool feature. https://www.bennadel.com/blog/4810-ask-ben-destructuring-complex-queries-into-coldfusion-objects.htm

I'm upgrading an old app from 1.4 -> 2.5.1 but jumping to lucee 6 at the same time. I've hit a few issues on application start - has any testing been done on the 2.5.x branch against lucee 6, or shall I just try and jump to 3.x ?

This is a small UDF that I use to dump-out model without the hundreds of UDFs in the execution context -- https://www.bennadel.com/blog/4812-light-weight-dumpmodel-function-for-cfwheels-in-coldfusion.htm

This is a fairly Lucee-specific issues (since the

showUDFs=false

flag doesn't actually omit functions from teh

CFDump

/

dump()

output). But, has been very helpful for me, especially since it works recursively for models in models in models.

A small article on how I'm working around a bug in the way the CFWheels ORM handles nested resources that are name-spaced: https://www.bennadel.com/blog/4813-working-around-sql-namespace-bug-in-cfwheels-models-in-coldfusion.htm

Just some Friday Fluff

You've certainly been busy 🙂

Very cool

Very cool! I still remember the conversation that brought CFWheels back—reviving it from the dead: https://groups.google.com/g/cfwheels/c/vhTZab-bplY/m/Vp1wI1OFK_gJ (Major props to @neokoenig for leading the effort)

ChatGPT really does make up whatever it wants huh.

I've been playing around with this and the pattern that I have found gives the best results is if you can figure out a way to get the ai you are using to test itself. I'm using Claude Code for this. I ask it to do something, then have it use puppeteer and commandbox and wheels cli to test what it did and "iterate till it works" then I tell it to update it's memory with /init. This way as it learns to do more and more it keeps adding that to its memory. Does it still hallucinate and come up with code that may be perfectly valid in another context, sure it does. But getting it to go through its four, five, ten iterations on it's own is what has yielded the best results.

If anyone is curious, an article about using "Service" objects (an exploration) in Wheels -- https://www.bennadel.com/blog/4819-creating-service-objects-in-the-cfwheels-coldfusion-framework.htm

In the spirit of ChatGPT just making stuff up 😄

Here's a quick look at one way to use Custom Tags in Wheels without giving up all of the wheels mechanics: https://www.bennadel.com/blog/4820-using-coldfusion-custom-tags-and-modules-in-cfwheels.htm

I would love to have some sort of unified API for the property + validation stuff. Even if it was just a proxy to the other validation methods. Like:

define(
	property = "name",
	required = true,
	minLength = 0, // can be blank.
	maxLength = 100
);
define(
	property = "salary",
	required = false,
	minValue = 0,
	maxValue = nullValue(),
	onlyIntegers = false
);

... anyone ever try something like this?

Also, another thing I've been thinking about: Iterating Business Objects (IBO). This is basically the notion that a single CFC would wrap a record-set internally. And have methods like

.hasNext()

and

.next()

, and would push a row of data into the CFC properties during the iteration. Basically, it would safe on the cost of instantiating a bunch of CFCs for recrord-sets. Maybe there could be a

returnAs="ibo"

...... just thinking out loud. This might be something I toy with to see what it looks / feels like. The main value would be to take advantage of CFC computed methods and association even without having to have a CFC-per-row.

For anyone curious, I created a

Mapper.cfc

which wraps the

mapper()

API and makes a more explicit extension point (for my own convenience methods): https://www.bennadel.com/blog/4823-customizing-the-router-in-the-cfwheels-coldfusion-framework.htm

Some more digging into the Wheels internals, and I realized I was able to override the

$selectClause()

method to monkey-patch a bug fix.

At work, we use the #C06TT5M7V #ColdFusion framework and the Microsoft #C082RPGTZ Server database. In a previous post, I talked about a bug in Wheels that was turning namespaced Models into syntactically invalid MSSQL statements. At the time, I was working around that bug by explicitly writing my

SELECT

statements (instead of letting the Object Relational Mapper, ORM, write them for me). Last week, however, I realized that I could (dangerously) override one of the Wheels internal methods -

$selectClause()

- and apply the fix globally within my #CFML application. Read more: https://bennadel.com/4813#update-overriding-selectclause

glad to see the traction this project has gained.. looking forward to getting into it. i've been using other frameworks for years so this will be a learning curve, but excited to see what it has to offer.

This isn't Wheels-specific, but it's Wheels-inspired. Basically, I want all the goodness of the global functions but with a little less magic 🪄 : > At PAI Industries Inc. we use the #CFWheels #ColdFusion framework. Wheels has a lot of "magic". I love the fact that I can use

e()

as an alias for

encodeForHtml()

in any of the #CFML or #CFC templates in our application. But, I don't love that functions just show up out of thin air. I'm wondering if I can balance the magic with the convenience by exposing runtime extensions and polyfills using a CFInclude / CFModule mechanic. This would make things more explicit; but, also keep things super simple. Read more: https://bennadel.com/4825

I think it would be cool to have two layers of validation: • Validation that should always run. • Validation that should only run if there are no error messages. This would allow all the "simple" validation (is empty, is non-zero) to take place. Then, it would allow other complex validations to run only if none of the simple ones failed (ex, uniqueness checks).

Kind of like what

unless

might do:

validatesNumericalityOf( property = "userID" greaterthan = 0, unless = "hasErrors()" )

Aside - I don't actually know how

unless

works.

It feels like the

allowBlank

arg is trying to work around this issue. But, I think that would become irrelevant if you had two-phase validation. .... just thinking out loud.

Is there a plan for a stable release of 3.x soon? I’m finding quite a few issues with the snapshots