Wow – UserLand plans open-source release of Frontier kernel. The kernel includes things like the UserTalk script interpreter, all the UserTalk builtin functions, the Frontier ODB and the Frontier web server.
I don't expect that it will gather as many developers as the big open source scripting languages, but it would be a worthy addition to the community. It used to be cutting-edge, and the system still has a bunch of great ideas, despite having been surpassed in various areas (speed, reliability, debuggability, popularity) by other competitors. Enough to be worth saving, for sure.
Just off the top of my head, I can think of some interesting projects to do with the Frontier source:
– remove the dependency on the GUI, so it could be run as a Windows service on NT/2K/XP (as modern server software does).
– once that's done, a UNIX port might not be so hard. Presumably it's fairly portable already, as right now it runs on two different versions of Mac OS as well as Windows.
– perhaps after those two, or perhaps first, someone could rip out UserTalk and make a standalone interpreter, so you can use it the way you might use Perl or Python at the moment.
– this is a bit of a long shot, but if you could make the UserTalk interpreter support the Python module API or something similar, a whole bunch of open source libraries (DB access, etc) would suddenly be within reach.
Anyway, that's all for now. Announcements of open-source releases usually precede the actual releases by quite a while, so don't hold your breath. When this release actually happens, though, it will be a lot of fun 🙂
OK – I'll say it.
Dave Winer and Frontier were my main introductions to the web, CMS, webapps, personal publishing and open standards. Without Dave – I wouldn't be sitting here today.
I can remember Dave telling me about this guy Eric Raymond – and his Cathedral white paper. It really pissed off Dave. He had made his Frontier tool available – for free – in the early 90's and had had direct experience in dealing with people expecting a whole lot for free. Yet Dave knew that by having a strong developer community behind him – that's all that mattered.
So Dave spent a few years trying to figure this all out. He then made the decision that he HAD to charge for Frontier – or else he'd never get corporate uptake. This is just as Linux and all the open source stuff was starting.
So now jump to years later. Frontier is finally open source – but it's too late. Php and python came along, Zope, Drupal – all sorts of stuff that does exactly what Frontier and….. well let me correct myself – not EXACTLY what Frontier does – as it STILL is a revolutionary IDE – as the integrated outliner makes o-o programming actually culpable.
But there are still problems with Frontier – memory leaks aside. And it sure seems like it's five years too late. That doesn't mean folks should'nt use it. My buddy Paolo uses it for his eVectors knowledge management stuff. Radio is built on top of Friontier and gives unprecedented extensibility that I'm sure many a MT user wishes they had.
I spent years pitching Frontier as the backend solution to our front-end needs. Most people would look at me and say “Huh?” But I'm proud of Dave and Frontier and I hope that LOTS of college kids give it a try! [Marc's Voice]