Even though I am really looking forward to Adobe MAX in Las Vegas this fall (probably because I love BlackJack and Vegas), CFUnited is where "my peeps" are at since it is focused almost 100% on ColdFusion. Through some quirk of fate, I am not giving a session this year (I guess I am not famous enough :-). But I will be hanging at the Adobe booth for large chunks of time and attending as many sessions as I can. I really find it fascinating to hear people explain features of CF that I have had a hand in designing. It is almost like a game of telephone - where someone whispers a message to another person on down the line, then you compare what the last person in the line says to the original message. The result is always surprising. Sometimes the message gets through, sometimes it gets really garbled. :-) But in every case it allows me to get a better view of the problems our customers are solving (or not solving) with CF so that next time I can get it closer to right.
This is your chance to harass me about how CFCs should be "just like" Java objects, why CFML should be strongly typed, why we should rewrite the file browser in the CF administrator to not use Java, why you can't possibly use CF anymore because we don't have {cfimap,cfimage,etc, etc}
See you there!
4 comments:
Hi Tom,
Since you asked, I'll bug you in person :->
In the meantime, here's my shortlist . . .
If anyone desperately needs strong typing, interfaces and nulls, feel free to throw them in, but PLEASE make them optional (or leave them out entirely). Some of us don't work in big teams and like to leverage the flexibility of dynamic typing.
On the documentation front, could somebody get somebody to write up a short white paper on the performance implications of using return types and cfargument types as I've heard that this actually hurts performance and would like to understand the implications of that a little better.
Finally, if you have to look at other languages, please look at Ruby more than Java. Ruby allows hackers to be productive. ColdFusion should allow n00bs and hackers alike to be productive. Java just makes for lots of typing and unproductivity whatever your experience level. I just don't see ColdFusion being the development language of choice for projects requiring 20 coders to collaborate, so leave that to Java and help us 1-5 person teams to be more productive!
If there was one personal wish-list item I might mention in person, finish what you started with cfscript. It decreases the signal to noise ratio of characters that have to be there to express programming intent vs. characters required to keep the parser happy. Feel free to leave CF standard for those who prefer, but at the very least wrap every cf only function (cffile, cftry, etc.) with a script accessible call so we don't have to keep doing that ourselves in our code and getting the productivity and performance hit it induces. Ideally even allow methods, arguments and returns to be done using a cfscript syntax and for those files don't require the cfscript tags at all - perhaps a cfs file extension? I guess we could go and write a preprocessing generator to turn script style files into cf classic, but it'd make life much easier if you put it in the box. That way we could attract more hackers and get all of our productivity up and our carpal tunnel index down a little!
Knew you shouldn't have asked what we wanted!!!
See you at CF United
Best Wishes,
Peter
I will note that I did not actually ask for CF wishes in the above post. :-)
Also, unless you test the performance in *your* application, don't avoid using CFML language features just because you "heard" they are slow. A lot of that is just misunderstands or plain wrong.
Hi Tom,
Apologies for that. Like most people, I guess I read what I'd like to see, not what the author actually wrote :->
The performance comment came from Joe Rinehart via Sean Corfield and included actual performance numbers for a real application (http://clearsoftware.net/index.cfm?mode=entry&entry=FB980E37-3048-55C9-43319854C6D8073D).
I would love one day to find a way of understanding the approaches and algorithms used so there was a set of agreed upon performance heuristics (if you're doing x, you may want to consider y) rather than having to build one (for performance testing) to throw away!
Thanks for all the amazing work. See you at CF United.
Best Wishes,
Peter
Speaking of telephone, I was witness to some local ISP tech, explain how Ethernet worked to Bob Metcalf and why what Bob wanted, couldn't be done.
Some just never learn.
Post a Comment