News and commentary from the cross-platform scripting community.
|
||
|
Mail Starting 10/20/97 | |
CORBA and RMI are very different.
From: mike@wellsfargo.com (Mike Brodhead);
Sent at 10/20/97; 2:17:00 PM;
factual error?Both are allow programs to talk to one another over the net. CORBA allows components written in different languages on different platforms to invoke one another. It is championed by the Object Management group.
RMI is strictly a Java-to-Java affair. Because the same language is on both ends of the connection, RMI has a few features that CORBA does not.
I'm not advocating one technique over the other. (At least not in *this* message (grin).) I'd just like to make it clear that RMI and CORBA are not the same thing.
--Mike Brodhead
Internal Webmaster
Wells Fargo & Co.PS: If you are interested in learning more about CORBA vs. RMI, _Advanced_Java_Networking_ by Prashant Sridharan (ISBN 0-13-749136-0) provides an excellent introduction.
While I don't know of any Java applets using RMI, I know that there are Java applications on the web that use RMI. Furthermore, the one that I helped write, was started 10 months ago and finished in May, so I can only presume that at least some of "the RMI interfaces were finished before Microsoft froze IE 4.0." OTOH, I don't know if/how they have changed over the last 5 months.
From: adam@student.net (Adam Trachtenberg);
Sent at 10/20/97; 4:44:26 PM;
Java & RMIOne of the reasons we wrote an application instead of an applet was because no web browsers supported RMI. Still, it was very nice that of the 4 people in the group, different people could develop and classes on their operating systems of choice: MacOS, Win 95/NT, and Solaris and have them work as expected when they were integrated on Solaris. I certainly would have been more than a little angry to find out that something broke because a certain platform's JVM defined extra or different methods within the java.rmi (or java.awt, etc.) classes and the person who programmed that object had unknowingly used one of those methods. Likewise, I would have liked to hope that if I had chosen to write an applet using RMI, that all the web browsers that promised to implement RMI would do so according to the published public specification; if everyone could fiddle around with the standard java classes, what would be the point of learning the language because every platform would have a different dialect?
From my point of view as a person using Java, I don't care if Microsoft (or Steve Jobs for Rhapsody) defines more efficient classes for their platform. Without a doubt, current JVMs are slow and I welcome any and all faster ones. But, don't redefine or omit classes! I don't want users to find out that my program "broke" because they changed their JVM -- or web browser. Now, Netscape might not "pass the test," but I don't equate "not having implemented" with "decided to implement differently." If Netscape does the same thing, I'd be just as frustrated.
http://www.trachtenberg.com/fish/
Straight from the FCC:
From: cameron@michweb.net (Cameron Barrett);
Sent at 10/20/97; 4:24:36 PM;
Everything You Ever Wanted to Know About 800/888http://www.fcc.gov/Bureaus/Common_Carrier/Notices/fcc95419.html
There's a FAQ (albeit kinda hard to read) about 800/888 numbers too:
http://www.fcc.gov/ccb/888/888faq.html
Stayin' alive...
From: rtmd30@email.sps.mot.com (rtmd30);
Sent at 10/20/97; 1:22:50 PM;
Re:The Man Behind the Curtain> Last night, for example, I posted informal performance results of
> our not-yet-released Windows software as compared to our Mac
> software. Under some circumstances the Windows software is much
> faster. A frequent correspondent writes saying that I'm biased and
> that something insidious is going on. Jeez! No way. Sure I want the
> Windows software to be faster. The Mac is too slow. Macs may not get
> faster. Frontier/Win is my new thing. I want it to be great and I want
> people to love it. Is that insidious? No way.This is an issue we're grappling with here.
While there are lots of manager types who've jumped to recommending NT, without showing just cause, us Mac support folks who've used NT, are having to figure out why NT blows the Mac's IP out the door. RAW speed goes to the Intel/Windows/NT world.
Sophistication in the user environment goes to Apple. I'm afraid that sophistication isn't enough anymore though. Stability and speed and sophistication are the winning combo, and the OS/platform that gets two of three will capture the mind share, and eventually the pocket book.
Operating a bulldozer is looking like a better profession than supporting Macs day by day. :-)
This entire saga of Java and Microsoft reminds me of the conflict that occurred when Christianity ran up against all of the pagan religions. In the New World and even when Christianity spread into northern Europe, the missionaries wanted the natives to practice their religion their way. Inevitably what you got was a "sincerity" religion.
From: Greg_Kucharo@NeTpower.com (Greg Kucharo);
Sent at 10/20/97; 1:03:49 PM;
Man behind the curtainThe pagans merged some of the Christian beliefs with their own well developed local religions. This is how we have holidays like Christmas with a paganesque deity like Santa Claus. Even Holloween is derived from pagan rituals. Christmas was never actually celebrated as a holiday until the European pagans absorbed the celebration of the birth of Christ in the Winter Solstice.
The pagans didn't really see a reason to stop worshipping the gods they had had for thousands of years. Only after a show of force and a mild interest did they kowtow to the missionaries. Of course the Christians we're appalled by the pagan gods, they felt that theirs was the one true way.
Religion destroyed Apple. If Sun doesn't watch out, it will destroy them as well. People will always bend the environment to their own liking. It's just human nature.
My roommate recently got set up to do credit checks (for his real estate venture) and during some discussions with the credit company, learned some interesting things.
From: hls@uac.com (Herb Singleton);
Sent at 10/20/97; 3:41:00 PM;
Man Behind the CurtainOne of the things he learned was that 1-800 numbers can and *do* sell your name and number to telemarketing companies. After all, the companies with 800 numbers need to have access to your name and number so they can be billed for the call.
Most people aren't aware of the fact that when they call an 800 number, the phone number they're calling from is readily available to the party operating the 800 number.
From: joelm@eskimo.com (Joel McNamara);
Sent at 10/20/97; 12:33:59 PM;
Re:The Man Behind the CurtainWith CD-ROM and Net reverse directories, it's trivial to then look up the name and address associated with the originating number, and viola, you're suddenly on another junk mail or telephone solicitation list.
It's little known technical gotchas like this that contribute to our eroding levels of privacy.
DoJ is dangerously close to codifying an old software paradigm of applications walled off from operating systems. As you have pointed out, the Mac's interoperability began chipping away at that, and the possibility of NCs running a browser only should prove that the Chinese wall is no wall at all. I'm not saying that I want their "seamless integration". But if I don't want it, that will contribute to my decision to not buy Microsoft's operating system. Imagine being barred by consent degree from offering both communications technology and information services . . . oh wait, that happened four decades ago with IBM.
From: btanen@tiac.net (Ben Tanen);
Sent at 10/20/97; 2:53:56 PM;
DoJJohn:
From: dwiner@well.com (Dave Winer);
Re:The Man Behind the CurtainI understand that the RMI classes can be downloaded from microsoft.com.
Why didn't Sun take the high road and choose COM as their underpinnings for interapplication communi cation?
Were the RMI interfaces finished when Microsoft froze IE 4.0?
Are any Java applets on the net using RMI?
Does Netscape pass the test? Does that matter?
What kind of communication was there between MS and Sun while all this was going on? Was Sun coopera ting with MS?
It isn't Microsoft's job to organize the Unix community.
Dave
Sun isn't saying that Java can only be used to implement CORBA. Sun is happy that Microsoft has its own proprietary method for invoking remote methods. What Sun is unhappy about is that Microsoft *deleted* the standard way of invoking remote methods -- the one that's supposed to work everywhere.
From: gnu@toad.com (John Gilmore);
Sent at 10/20/97; 11:50:05 AM;
Re:The Man Behind the CurtainIt would cost Microsoft nothing technically to implement the standard method in addition to their proprietary method. They already have the code, and it already works. But they deleted it.
There are no boring subjects. Only bored people. Billions of industry dollars are at risk on this "plumbing protocol" issue. I'm not bored.
John
PS: Unix lost the API wars to Microsoft because the Unix companies got tied up in boring arguments about plumbing protocols, rather than settling on a single API that they all could ship. With Java they're finally settling on a single API. And now who's coming along and making different versions of the API?
Actually, it was a very tough choice, but it was a choice that had to be made. The targets for Java include settop boxes, credit cards, etc. We cannot allow the core set of Java classes to balloon out of control. An OS has this kind of leeway, because bigger storage devices are always available, and who cares about another 35Mb app on a 2.5 Gig disk?
From: xtian@Eng.Sun.COM (xtian);
Sent at 10/20/97; 11:23:00 AM;
Re:The Man Behind the CurtainJava is targeting a lot of devices with much smaller footprints than a desktop PC. And, Java is an _addition_ to the underlying OS...so we cannot allow it to just grow indiscriminately. Some choices need to be made (CORBA vs COM, JFC vs IFC vs AFC, Security Model A vs Security Model B, etc, etc, etc.).
Ideally we would want to support all of these things and not exclude anything, but that is not reasonable.
So CORBA won out over COM. It had wider support from a wider group of platform vendors. It is not an arbitrary limitation.
I don't mind so much that Microsoft may choose to include extra classes in its VM or J/Direct, or wh atever. So it's fine that they give developers the ability to use COM.
From: seitz@mail.medscape.com (Bill Seitz);
Sent at 10/20/97; 1:21:01 PM;
Microsoft supporting RMIOn the other hand, there's no particular reason for them NOT to include the RMI classes, other than the fact that MS wants to impose COM on everyone. (They talk about not bloating up software with unn ecessary stuff, but who are they kidding?)
Yes, Java is a brand. It is intended to communicate certain things to its consumers. One thing it tr ies to communicate is WORA. Since interapp communication is important to this, being able to assume a certain plumbing protocol seems important (to me). And I'm not sure anyone could hide the differen ce between COM and RMI from developers without making some lowest-common-denominator choices. So I'm not sure how MS's refusal to ship RMI does not count as "standing in the way".
The Romans must have thought: 'Our Empire is here to stay'. Probably the Soviets did the same. So when you wrote: 'Microsoft is here to stay' (in The Man Behind the Curtain), what timeframe did you have in mind? And does that include the possibility of Bill Gates no longer running the company (old age, career switch, accident, move to White House)?
From: beno@xs4all.nl (Michel Benevento);
Sent at 10/20/97; 7:23:45 PM;
Panta rheiPanta rhei -- everything flows, nothing lasts. Especially in a world where mindboggling change seems the only constant. Once Everything runs on MS software, the only way they can keep on flowing is southward.
PS: I fully understand why moving to W32 right now is a smart move. I just hate to hear you speaking of Frontier/Mac in the past tense.
Your column today brought me back to the last time I had a decent conversation with Bill Gates.
From: marc@canter.com (Marc Canter);
Sent at 10/20/97; 9:16:42 AM;
Re:The Man Behind the CurtainWe were both at some conference and we were standing around a table - doing what Stewert Alsop calls the one required activity of any conference - eating large crab legs and shrimp.
He asked me how things were going at Macromedia - and how he thought Director was such a powerful tool. I thanked him and reminded him that Microsoft passed on publishing the product - back in 1984 (they guy who refused to show it to Gates was fired soon there after!)
We babbled on about kids (I had them, he wanted some) and about how making software was VERY similar to raising children -how at one point you have to let them go and grow on their own.
You can only hope that you've instilled enough in them to grow into good people.
Anyway the one point that stuck in my mind was that he didn't see why anyone would want to develop WORA software. He just didn't see the point. "Everything is slower and besides, why would you want to develop for anything besides Windows?"
I was completely opposed to this notion - as our Director player was connecting together the Mac and Windows and how it made sense to multimedia devlopers - to NOT have to repeat their work so that they could amortize their investments in content and interactive user interfaces.
I like to think that Director and it's run-time player - now called Shockwave (along with Postscript) was one of the building blocks of cross platform authoring, and that it made the world a better place.
I realized that at the time -Gates just didn't care that much about multimedia develoeprs, and that all he thought was important was X number of Visual Basic programmers, creating Y number of applications for Z number of corporate customers.
Jump forward 6 years to the Java licensing deal. When I heard that Microsoft had licensed Java - I immediately thought "Wow how Bill has changed his song." The real issue of Microsoft jumping on the Internet bandwagon had nothing to do with making Excel or Word work on the internet (they would have been done that anyway) - it was "Microsoft supporting cross platform WORA software!" Wow! Now THAT'S a change!
Gates has gone from ignoring multimedia develoeprs, to courting them and creating his own in-house staff to ignoring them again and focusing on web developers this time around. Same song, different key.
I've watched this Internet thing unfold, and the amazing things possible with an integrated HTML browser built into the OS, I now realize that Gates HAS NOT changed.
He still doesn't see the sense of developing for anything but Windows. It's similar to my approach to developing with Director - why would you use anything else?
Or Dave's approach to scripting - clearly Frontier is HIS platform of choice.
So what's the conclusion I reach from this recent flurry of activity and conversations? That we're all three software guys - all three about the same age (I'm actually the youngest - being a babe of 40) - all three raised in a world of FUD, chaos and a focus on the future.
There has never been a time in our adult lives that we haven't had some burr under out butts, focused on getting our software out there and accepted by our limited, but influential niche of customers (granted Gates' niche is a little larger than Frontier's or Director's!)
And we don't follow the price of our stocks. That's not why we started our companies or do what we do.
Because we can make a change and help make the world a better place, puts us into a unique place in the world order. Even Scott McNealy wants to help the world - and make a change.
Now the only issue is - do you do it with CORBA or COM? I say both.
I just read your piece on Bill Gates. I just wanted to say that I think Sun is internally conflicted about CORBA as well. RMI is not CORBA based, but is its own, roll-your-own RPC mechanism that's very Java friendly. It has some flaws, however, but is still very Java friendly.
From: jpayne@marimba.com (Jonathan Payne);
Sent at 10/20/97; 10:15:18 AM;
Re:The Man Behind the CurtainCORBA is not as Java friendly in that you need either new syntax or some ugly methods to perform certain operations.
CORBA and RMI are different from COM.
That's my understanding.
Your pieces on Cadillac Desert & New Orleans hit a nerve.
From: danlyke@flutterby.com (Dan Lyke);
Sent at 10/20/97; 6:54:19 AM;
Cadillac Desert & New OrleansI moved to California a little over two years ago from Chattanooga Tennessee.
In August I went to the Sierras, spent a week away from Marin, and I realized that I'm not cut out for the SF Bay area. I can assuage my guilt by rationalizing that where I live takes its water from the Russian River, and not from a dammed valley that once was the equal of Yosemite (Hetch-Hetchy), but I miss the scenery, the whitewater, the rock climbing, the solitude.
I'm wasting an hour a day or more in traffic, and if I make the concessions to the environment and take the bus (there's no bike route across the Richmond-San Rafael bridge) that quickly goes to two hours. And compared to many in Silicon Valley I've got it easy.
And I'd far rather be buying up and preserving large tracts of land than spending outrageous amounts for cheaply built houses in the middle of fire country.
While it's obvious (to me, at least) that while the real solution to some of the environmental issues we're facing is to at least go for zero population growth, a first step might be to work towards not growing our huge population centers in the middle of the desert and then importing the lacking resources at the expense of ecologies hundreds of miles around.
This isn't to say that there aren't advantages for developments such as LA. Concentrating the types of people who actually like smog-filled valleys, non-native foliage and awful sprawling suburbs near fault lines is a good thing.
But I think that with a little social change you can party in New Orleans and I can give up my commute and have a garden, and that people like me don't have to abandon their communities and move to the desert in order to have a career.
So what do we have to do to make telecommuting more socially acceptable? I don't enjoy constantly selling myself enough to do the freelance stuff (did that for a couple of years, may end up doing that again), but surely there are managers and producers who can see past their fear of loss of control and look towards reduced costs of living, (according to today's SF Chronicle) 20% greater productivity, and reduced capital costs because I wouldn't have all the duplicated equipment between home and work (as I'm sure most programmers do)?
Or maybe I should just get off my duff and go back to the freelancing and entrepreneuring.
Plumbing is an implementation detail. If I can use the RMI API, and behind the scenes the messages are sent using COM, then that's great -- I reap the benefit of a performance boost on Windows. That way the Win32 platform can compete in the market on value (high performance for Java apps) not on sneaky bundling tricks.
From: alex@stinky.com (Alex Chaffee);
Sent at 10/20/97; 12:17:15 PM;
Re:The Man Behind the CurtainBut that's not how Microsoft is doing things :-(
I'm sure others have reported this back to you, but an 800 number is not necessarily toll-free. Scam artists have wired up 800 numbers that redirect to destinations outside the U.S., and the toll charges get back to the caller.
From: dannyg@dannyg.com (Danny Goodman);
Sent at 10/20/97; 9:01:12 AM;
Re:The Man Behind the CurtainI learned recently that there is no actual rule or law that says 800 (and now 888) numbers are free. They in fact COULD BE toll numbers, although by popular convention they in reality have not been until now. So be a bit more careful here before making such blanket statements to all your listeners!
From: yvon@cais.com ();
Sent at 10/20/97; 11:40:57 AM;
Re:The Man Behind the CurtainRMI is *NOT* about plumbing. It's about API. The RMI API is clean, simple, and totally Java-friendly. This weekend I knocked out a multithreaded chat server using RMI in about 4 hours, no lie. I find the API remarkably easy to use.
From: alex@stinky.com (Alex Chaffee);
Sent at 10/20/97; 11:32:57 AM;
Re:The Man Behind the CurtainAnd Microsoft doesn't want me to use it.
They don't ship the RMI classes with their Java products. They want people to use COM instead, which still doesn't work cross-platform. If I'd used COM to write my chat program, then the server would only run on Windows, and the client applet would only run on Windows.
*That's* trying to control the market.