ColdFusion Needs A Roadmap
I've been actively using ColdFusion since early 1997 (side note: I almost had a coronary as I just thought about that). In all that time, though, there has never (publicly) been any sort of development roadmap for the product. All we've been able to do is cross our fingers and hope that Allaire/Macromedia/Adobe finds it in their heart to add or support a particular feature. Sure, Ben has made the CFUG rounds before each ColdFusion release to generate some buzz and gather feedback, but that's not really what I'm looking for.
As a developer who also has a significant hand in my company's IT procurement processes, there are certain things I need to know about any product before I'll move ahead with them. Our larger purchases are not "spur of the moment." (In this context, "larger purchases" means anything over a couple thousand dollars.) However, the lack of a roadmap really makes it hard for me to properly plan our infrastructure and development timelines. Don't misunderstand, I'm not looking for Adobe to announce every new/enhanced/fixed feature of the product in advance of its release. As a developer, I certainly know that planned features are always constrained by time and resources and are subject to falling out (see replicating session-based CFCs in CFMX 7 as an example ;). However, here are a few concrete areas where the lack of a roadmap in a hindrance:
- 64-bit support
- Java 1.5 support
- Support for the latest Sun 1.4 JDK (1.4.2_11 is currently the latest JDK release, but 1.4.2_09 is the latest officially supported by Adobe)
- Tentative Scorpio (aka ColdFusion 8) product cycle
Why do I need answers on the above? For starters, we're imminently purchasing a couple of new servers for our web site. In working with my sys admin, he's asking basic questions that I simply need to know the answers to in order to formulate an accurate spec. He's asking why we wouldn't go with 64-bit servers. The only answer I can give right now is that it's because Adobe doesn't seem to have plans to get ColdFusion working on 64-bit platforms (yes, I know it'll run in 32-bit compatibility mode, but that's not the same), so why spend the extra money? Do I necessarily believe Adobe hase no internal plans for this support? Not at all. However, in the absence of any other publicly available information, that's the only conclusion I can draw at this point. It could come tomorrow, it could come next year for all I know.
Regarding the Scorpio product cycle, it would be nice to have some sort of feel for the timeline. Is an alpha/beta imminent? Since Mystic is apparently going to be ColdFusion MX 7.02, does that push out the Scorpio cycle? With the new servers we're buying, I am also planning to upgrade our site to CFMX 7. However, if Scorpio is on its way in the short-term, maybe I'll hold off and take advantage of its new features. There's nothing on my company's web site that's screaming for a specific CFMX 7 feature. However, with Scorpio's rumored new resource monitoring features, that's something I could use and maybe I'll hold off on a purchase of FusionReactor or SeeFusion.
There's been frequent mention in the blogosphere lately about how Adobe and Microsoft are seemingly on their way for a competitive showdown. I don't know if I necessarily believe that. However, if it's true, one area that I believe Microsoft has a significant leg up is with their roadmaps. Sure, you can insert the latest Vista delay joke here, but the reality is that I know where their products are going. From Vista to Visual Studio to SQL Server, I know when I can roughly expect service packs, upgrades, etc. I can't say that about any product at Adobe other than Flex 2. I hope that Adobe learns from the (in my opinion) success of the Flex 2 public beta. Since I knew that some time in June is the expected release, I have been able to schedule development cycles with it accordingly. I just wish I could do the same with ColdFusion...
 |
 |
 |
 |
|
|
 |
 |
 |
 |
http://corfield.org/blog/index.cfm/do/blog.entry/e...
But in general, I agree that some kind of public road map wouldn't hurt.
This is something that has really been missing for years in the CF community. Personally I feel that we are kept in the dark 90% of time. The only light we get are Ben's little teasing that he comes out with every once in a while.
First item that should be added to the roadmap:
Move from J2EE to .Net runtime.
Tony, I'm not sure about _moving_ from J2EE to .NET, but I wonder what the interest level is from existing Adobe customers as far as CFML running on .NET. Obviously, New Atlanta has BlueDragon.NET already and I wonder if that niche market really needs multiple players?
1. What ends up in each product release of ColdFusion seems to be very customer-driven. Ben (and Damon) talk about this on their blogs at various points. When planning a new release, often two years in advance of the launch date, the team go out and talk to customers about what they want solved and that's mostly what drives that release. Once the themes and major goals for a release are nailed down, Ben and others start mentioning them publicly to get more feedback. As Ben has repeatedly said in the past, features don't always make it into a release for a number of reasons. Publishing an official roadmap saying feature X is coming in release Y on date Z is a pretty dangerous thing to do if you can't be certain of that - suppose you plan your applications, hardware and schedule around that roadmap and the product team have to change things? You'd be pissed. And that brings me to my second point...
2. You mention Microsoft roadmaps and they fall into exactly the situation that I raise in point #1. They've published features and dates and look what has happened: Vista has slipped repeatedly, it's had features gutted from it, most everything that was promised has now either morphed, disappeared or been rescheduled. I talk to a lot of software vendors (part of my job is technology selection) and of course I always ask about product roadmaps but it's very rare they can tell me much of their plans beyond the next scheduled release, and even then they tend to be pretty vague and won't promise anything specific.
A couple of times in the past, I've relied on a vendor's promise for "the next release" and been let down - even when that release has been pretty close to shipping! So I'm very wary now of roadmaps.
To me, there's a distinction between "feature" roadmaps and "platform" roadmaps. The feature roadmaps (i.e., instance monitoring, event gateways, etc.) is _not_ what I'm looking for, nor would I expect that. However, platform roadmaps (i.e., 64-bit support, JDK 1.5 support, etc.) is definitely something I'd like to see. I'm not looking for a specific date, but even something along the lines of "we are targeting 64-bit support for Q4 2006." This gives Adobe the flexibility of releasing something within a three month range while allowing me to plan out my hardware purchases now.
Again, I refer to my Flex 2 example. It was noted for a long time that it would ship "in the first half of 2006." That's great. I don't care about the _exact_ date. I just like knowing that Adobe's internal timelines have the end of June as the realistic upper bound on the delivery estimate, and I can plan out my own Flex 2 development efforts accordingly.
Believe me, I hear you on the "vendor's promise" letdown. I've been very up front in the past that I was unbelievably disappointed when Ben promised at a CFUG meeting (during the CFMX 7 tour) that replicating session-scoped CFCs would make it in and it wound up dropping. Again though, that's the "feature" roadmap that I'm less interested in and I wouldn't hold Adobe as accountable for.
Obviusly there is an interest from Adobe customers for .Net runtimes for their products. Just take a look at the breeze presentation for Flex and it mentions that there are plans for a .Net runtime version of it. We need to face facts that .Net is the dominant platform for Windows development and espically with Win2003 and IIS6.0 performance enhancements, it only makes sense to get a .Net version of CF going. If Adobe doesn't want to spend the time to develope it (which I don't blame them), then they might want to look at buying out NewAtlanta to give them that edge. Like Sean mentions, with the products and features that MS kepts pushing back, it could be a big advantage for Adobe to push CF and Flex.
Personally I think Adobe should freeze all CF development and concentrate on a 7.1 release which fixes every single bug that's been reported. A stable and bug free release is more likely to sell the product than new features and more bugs. (Otherwise this will be the last version that we'll ever use.)
A public bug list would also be helpful. We must have wasted a week of development time with issues that boiled down to bugs that MMAdobe know about but haven't acknowledged or documented.
64-bit support: You can definitely run in a 64-bit VM with the pure-Java kit of CF, no problems. What do you want to run on?
Java 1.5 support: CF8's got it, and makes use of it if we're running on it for various things (although 1.4 is still supported in CF8, since many app servers are much slower than we are to jump to 1.5).
Support for the latest Sun 1.4 JDK (1.4.2_11 is currently the latest JDK release, but 1.4.2_09 is the latest officially supported by Adobe): 1.4.2_09 WAS the latest version when we started putting CF7.0.2 together :) Sun's since done some rapidfire updates, and we don't expect any issues with 1.4.2_XX, but 1.4.2_09 is what CF7.0.2 lays down, and it's tested with full QA regressios, all platforms, all configs, all tiers. Like we do, reviewing the delta big fix list in the JVM release notes each _XX release is a good habit...
Tentative Scorpio (aka ColdFusion 8) product cycle: Basic release roadmap: 7.0.2: soon. CF8: 2007 sometime. Likely an Updater release (or two?) between 7.0.2 and CF8 (good old bug fix releases you mention).
HTH
Damon
1) Make the proper decisions with regard to hardware procurements (I'm going with 64-bit now).
2) Know that the 1.4.2_11 JDK release doesn't blatantly break CF (aside: we're actually using it in our staging environment and it's been rock solid)
3) Commit to a CFMX 7 upgrade in production with the knowledge that your team is still looking to improve/enhance the product (vs. leaving it as is and moving on to CF 8)
4) Look forward to 2007!
To answer your 64-bit question, I'd like to be able to run JRun/ColdFusion on 64-bit Linux and know that the software is taking advantage of the larger per instance memory space that the 64-bit platform provides. JRun 4 with Updater 6 and JRun 4.5 support this platform, but I'd like to know that ColdFusion itself can take advantage of it.
Again, thanks for taking the time to leave these details. I know this is not the norm for your team (that's not a dig) and they really have helped me make a more informed long-term decision.