UserLand Planning to Release Frontier's Kernel

Dave announced this morning that UserLand is planning to release the source to Frontier's kernel sometime this year. There are still a lot of details to be worked out, and apparently they're doing it mainly to ensure the future of the platform: low expectations of community-building or bug fixing.

Years ago, Dave told me that if Apple released MacOS as open source, he'd release Frontier as open source. Well, Apple's kernel is now open source, and apparently Frontier's kernel will be, too. (That's just an interesting coincidence, though. This wasn't Dave's decision to make, now that he's only a board member and not running the company.)

Note that he didn't announce that Frontier will be open source: just Frontier's kernel. There's no indication that one will be able to download the source code and build the Frontier application. There's nothing that says anything to the contrary, either.

Months ago I caught a small hint that this was being considered, but there's been nothing since then and I completely forgot about it.

This news affects me and the software I write, directly and rather intensely. There are business decisions I've been trying to make about the future development path of some of Macobyte's software. If this new offering from UserLand is strong enough -- which I'm not counting on, by the way -- then my final decisions could be very different than what I thought they would be just a couple hours ago.

That's a big if, though, as I'd need to believe two things. First, that fixing bugs and adding features in the kernel would be a better use of Macrobyte's time than bringing my software to another platform. Second, that changes I make in the kernel source could somehow be put to immediate use in my software, rather than having to wait for a point release from UserLand.

What won't work at all is a situation where you still have to buy licenses to Frontier in order to do anything meaningful with the 'open source.' For example, if Frontier is updated so that replacement kernels can be dropped in easily, but you still have to buy the application, it's useless to me. I don't mind contributing improvements to the code, but I won't pay for that "privelege."

Whatever the outcome, this is the most interesting news to come out of UserLand in many years, and it gives me an option for the future of some of Macrobyte's software that I hadn't considered before today.

Update: Duncan has some comments, too. So does Daniel.

[Top]

They're Releasing it All!

Oh man. Dave clears up THE big question: they're releasing everything you need to build Frontier (the application) on both Windows and Macintosh.

Sheesh, that is not at all what I expected.

I need to go for a ride so I can think about the implications. I'm not yet sure if this changes anything, but it certainly could.

[Top]

RE: They're Releasing it All!

On 5/17/04, Thomas Winningham said:

>Congrats! I think its good, because it drives up the value of
>Conversant... it should surely cause more people to implement the
>product, which means more potential customers and converts to
>Conversant!

There's a good chance that this source-release is coming years too late, unfortunately.

Frankly, this move hasn't made my own decisions any easier. Quite the opposite.

[Top]

Re: They're Releasing it All!

Actually, no, I'm not looking at other CMS. That's not it at all. Can't say more right now, though.

[Top]

Scott Lawton, and More on Frontier's Release

Last Friday evening, Scott Lawton sent an email to a few of the Frontier gurus (who someone referred to as "expats"!) with a couple links:

• General Info
• His wish list

Scott's another old-hand around the Frontier ranch, and another who left to find work elsewhere. His comments have been very similar to others: this is cool, I won't be able to get involved, here's what I'd like to see happen, etc. That's a pretty good mirror of my own feelings.

His wish list includes syntax coloring in the outliner (and script editor): yeah, me too. He also asks for a regex search dialog, but that was added in 9.0.

He wants Python and Java integration, with Python as a first-class language. This sounds good, but also sounds like a huge amount of work. Would it be less work to integrate support for plugin language modules into Frontier, or build a new outliner for Eclipse based on the ideas we'll find in Frontier's outliner? I'm betting the latter.

Most of the rest of his ideas are fairly common, but I don't blame him at all for repeating them.

This isn't easy to say out loud, but I think what most of us want is Frontier's outliner, script debugger functionality, and object database in an otherwise-much-better IDE (like Eclipse), and the object database (and perhaps the language) available in a stable, high-performance server environment.

Frontier is still so... early 90's. I just don't know how or why we should expect it to be updated with 21st century features. What we really want (well, what *I* really want) is access to some of it's features in a modern system. The anachronisms in Frontier are found everywhere, and I've come to believe that updating the worst of them will involve almost rewriting it from the ground up.

I'm leaning towards Frontier's source becoming a place to learn more about object databases and outliners in order to improve other IDE's, rather than a strong platform offering that just needs a few improvements to catch up with the rest of what's out there. The alternative just doesn't seem realistic.

From what Dave has said, though, it looks like he'd be more than ok with that:

BTW, people who say that it's too late can't know that. And the expectations are low. All I want it to see the technology preserved. If one young programmer in Podunk learns something from it, the way I learned from reading the Unix kernel in the late 70s, then I'm happy. If it exists on a hard drive somewhere in the year 2040, that'll exceed my expectations wildly. If a bug gets fixed or a new technique is learned from reading the code, that would be fantastic. We're not finished giving, yet.

[Top]