...goes to Kevin Altis for pushing hard for Jython as a scripting language plugin for the Eclipse IDE.
...goes to Ticketmaster for a very plainspoken description of their use and relationship with OSS software and people.
All I know is what was presented, asked, and answered in the session. That said, they appear to use OSS to their advantage, manage it well, and give back to the community including hiring good OSS developers to implement features they need and that the community at large can also benefit from.
...goes to DJ Adams and Piers Harding for their servlet written in ABAP, the programming language inside SAP R/3.
Yesterday at the Applied XML conference an involved discussion took place about complex types, specifically a Map (aka Dictionary, associative array, property list). I was thinking about the origins of the problem more than the problem itself, because it didn't seem like such a problem in the first place.
Only in languages derived from C could Map be considered a complex type. (I translate "complex" to "unusual"). The major scripting languages (Perl, Python, Ruby) being discussed this week at OSCON have this type as the bread and butter of their existence. There is nothing complex about it, and you couldn't dream of connecting processes with these languages without passing Maps around.
The Great Ancestors of these languages, Lisp and Smalltalk, were created by the Gods themselves (MCarthy, Kay, et al.) fashioned out of the true primal matter, which is not
int, but association.
When Dave Winer froze XML-RPC (I don't know much of the details), it's most notable that it consists of two fundamental structuring capabilities: a list of things (called Array) and an association of things (called Struct).
No one argues over this. Every language has an XML-RPC implementation. It's funny that three or four years later the descendent, SOAP, is still wrestling with such a fundamental construct.
I think has more to do with ancestry than technology.
Attended Dana Moore's session on Python and other dynamic languages in a Java/dotnet world.
Fun demo manipulating the Microsoft clippy guy from Jython via XML-RPC to CPython manipulating the clippy guy's active X thinga-ma-thing.
All is love when it comes to dynamic languages.
November 2001 - Eclipse launch
Tools don't work well together. Tool vendors reinvent wheels. No vendor can do everything.
In Java no platform with strength of MSFT's Visual Studio.
Eclipse is a platform - extensible. Works out of the box. Companies contributed tools right away.
OSS chosen to support platform adoption.
"Tool stacks" per language type innefficient.
Eclipse is not a tool with an API. It is a platform for plug ins. Turns the picture upside down.
Small runtime + plugins. Plugins extend other plugins.
Architecture and demo (John Wiggin?)... internal team plays by same rules... plugins on small kernel... as all developers.
Team facility, repository support in OSS and commercial... (I guess this ties in to O'Relly's network collaboration talk)
Debug component... (who debugs programs?)
(Eclipse if you have not seen it is a Visual Studio look alike to my untrained eyes. Lots of panes with information here and there. People like these, I've not used them. Interesting challenge... I wonder how simple the UI could be made and still take advantage of useful plugins. How tied are the plugins to the UI look and feel? COuld you build Emacs in Eclipse?
(Talked with Kevin Altis last night... he's big on getting Jython into Eclipse as a scripting language. I've been doing a good bit of SWT programming in Jython outside of Eclipse... I think this would be a good idea and would lend toward getting an Emacs like capability into Eclipse, and could thus support a simpler interface Look and Feel for Emacs die hards at least. I'll look these guys up afterwards... this would be a good test of the plugin architecture, so maybe a few folks would like to collaborate (phrase of the day))
(Paul Buck is talking again...) COBOL development environment. Hades test framework. Eclipse modeling framework - abstract data types.
Technology PMC - further out, independent, experimental work, academics. "Coy" project (?) led by Instantiations (Portland!) doing something interesting... didn't say what. They're Smalltalk people, though and have done good tools stuff there.
New project for J2EE tools started last month looking for collaborators. See Eclipse.org
Next big step is moving beyond IDE into rich client platform. (Hmm.) SWT GUI toolkit for desktop Java applications.
(Hey - and desktop Jython applications!)
So new set of challenges... (well what are they?)
In closing... Websphere studio workbench is branded supported version from IBM. THE END
The value of Amazon, Google, etc. is the constant update via the people.
Network collaboration - Usenet is the mother.
Ad hocracy (Doctorow) - OSS is one.
Software paradigm of the future => *Collaboration* paradigm of the future.
Collabnet puts in OSS infrastructure inside organization's firewall, e.g. HP's printer drivers.
ASP.net reengineered ASP for XML - denied, backward compatibility. Six weeks time to hack it anyway. People took pieces of code here and there they liked then Gates annointed.
Network + Developers == OSS-like behavior.
"Listening to Napster" - keynote at a P2P conference. People can build a collective work as a by-product of their own self-interest.
Watching connections in Google, Amazon, elsewhere e.g. - we contribute with every link, book, etc.
Linus - essay in Open Sources - architecture of Linux and Internet supports individual contributions that add up.
Just now in throws of figuring out new business models. IBM Websphere, Mac OS X look like the Compaq business model. Selling commodity with add-in value.
Who will come along with the Dell model - "build commodity to suit." Key is the process of assembling the product, not what the product is.
Cable TV - pay per view not as successful as subscription. Aggregating larger packages of OSS.
Professional services business model not the most interesting thing.
UUNET made money with services - low margin ISP w/billions in revenue. Not by packaging UUNET software.
Network Solutions did not want DNS registration - then found the business model and did not want to give up the monopoly. Another ISP.
OSS developers need to pay attention to things not central... P2P, Rendezvous, Hydra, wireless, "social software"... all about network enabled collaboration. Or you will get behind the curve.
Increased customization - not just software. Look for hidden service business models. Collaboration beyond software.
Mapping software has not figured out use collaboration yet... location based services, individual contributors of data points into what's happening on the map.
Watch the alpha geeks - prefiguring larger movements. Screen scraping predicted web services. Rethink OSS in context of web services and network computing.
Who owns the data? License implications of data relationships. What will the bill or rights be for web services?
Kurzweil - long term thinking. Build out into the future. THE END
...so far goes to Kevin Altis for:
Note to those of you not used to the lawns west of the Rockies; there are no chiggers here, so sitting on the grass is fine :)
So true. I used to hate those damned itchy bastards. Which reminds me to stay of the grass in shorts when I go to Ohio in a couple of weeks.
From the referenced web page...
Probably no creature on earth can cause as much torment for its size than the tiny chigger... Their bites produce small, reddish welts on the skin accompanied by intense itching as irritating as acute cases of poison ivory or poison sumac. These symptoms often are the only way of learning that an outdoor area is infested since chiggers are so small that most cannot be seen without a magnifying glass.
Wireless is working, thanks O'Reilly, et al.
Attended the first couple hours of OSCAMP this afternoon.
1:30 The Economics of Open Source (Stormy Peters) 2:00 Linux in the Enterprise - An IBM Perspective (Dan Frye) 2:30 Legal Framework for Open Source (Larry Rosen)If you're reasonably up on Linux and OSS then the first two sessions were nothing new. These sessions are mainly for people and businesses just dipping their big toe in the water.
Dan Frye was asked (by me, actually) when the vendors will smooth out their differences. I ask because he made a big point about the hardware portability of Linux, but nothing about the differences among distros, which can be a big problem as well for businesses and individuals. And can result in hefty support fees, etc. because of vendor lock-in, which is not what OSS is about, is it?
Anyway his answer is progress is being made, it gets better over time, but no commitment to a timeline from the vendors.
Larry Rosen gave an interesting talk about OSS licenses and in particular his own license, which addresses concerns he has with the several dozen others that are available.
Came out last year at OSCON, but this is new to me.
Took a walk over to the Marriott to take care of some business... now back for the last hour of OSCAMP.
I will be showing a very simple start toward an adaptive blog tool at the Applied XML conference on Thursday, as an example of using XML, Python, and XPath.
The intersection between blogs, Wikis, semi-structured data, and more structured data is extremely rich. Something like Infopath is just scratching the surface, I'm sure.
I use OS X a lot and I am pleased that Apple by default installs usefull scripting languages like Python, Ruby, and Perl by default.
I use Jython with the Java Platform. It's a nice combination since a lot of "CPython" runs in Jython too, combining the two platforms with the ease of "scripting".
It really is a combination since Jython classes can be used in Java as well as the other way around.
I'd like to see this philosophy continue with dotnet as well. The scripting languages don't seem to be coming along as quickly. At least I don't see as much attention paid there. Smalltalk is coming along slowly with three implementations I know of.
Also I'm curious about the direction of SOAP, et al. Over time I would expect easier language integration across runtime engines and less attention paid to language integration within runtime engines.
This has to be easy and inexpensive to program to be truly effective. The hardware will make it fast.
If you use blogger, check your feed. My feed was turned off the last couple of days somehow, perhaps as part of their recent upgrade.
Anyway, I had to go back to the settings and turn it on. Everything seems fine.
Ted and others, re: OSCON...
I'm still interested in Patrick Logan's "What's Next" discussion. Can we do this Tuesday or Wednesday night?
Great. I am available Tuesday around 6:30. Wednesday is not as good. It looks like 6-7 or after 9.
If Sam is still interested, looks like he's not in town until Wednesday or Thursday.
James was interested too. Maybe he'll chime in.