News and commentary from the cross-platform scripting community.
Mail Starting 4/10/98
Sorry, a little late on this (still catching up)
From: email@example.com (Will Cate);
Sent at Fri, 10 Apr 1998 19:50:32 -0500;
Over the past 3 years, Prudential has been converting 58,000 desktops worldwide to Notes from whatever-they-were-using-before for mail.
Notes is a control-freak's dream-come-true: docu-databases with seven levels of access, link-ability, rich text & graphics support. It's not open. Verrry proprietary.
The scripting language is, of course, LotusScript, which I don't know much about, but appears to be essentially a knockoff of MS-BASIC.
Oh yes, and Notes Server has a built-in HTTP daemon ("Domino"), which converts Notes "documents" to HTML on-the-fly (this may mean it has something in common with efforts to make Frontier itself into an HTTP server? -- which, BTW, is a great idea, but I haven't followed the "Betty" thread very closely).
I actually don't care much for Notes. The Mac version is a dog; a RAM hog that installs about 250 separate files (like its PC counterpart). But the PC version runs acceptably. It generates a lot of support calls to our Help Desk. I don't think the web server works very well.
Outside of corp. America, I really can't see the appeal.
The interesting question might be, could you build a client-server messaging system, like Notes, on top of Frontier, where client-side scripts could refer to both local and server-side object databases? That'd be neat-o.
I've started to read the XML specs and related information and I don't know what kind of story I'm reading.
From: firstname.lastname@example.org (Andy Freeman);
Sent at Fri, 10 Apr 1998 14:54:36 -0700;
Yes, I know that XML is an interface/interchange definition, but is it one that requires lots of support? To put it another way, if I've got 5 people on my development team, how many of them must be XML experts? How much does it cost to play on the XML field?
There is also great interest in XML within the Director development community. We could use XML to describe and exchange data between Director apps or shockwave over the internet. Director has the hooks to make this happen right now. I haven't thought of a compelling example yet, but I will.
From: email@example.com (Ron Bearry);
Sent at Fri, 10 Apr 1998 14:34:03 -0700;
Re:XML Becomes Invisible
I think Michael is right, and his suggestions too. The question is the behavior of Frontier as an OLE server. It can be made to behave properly, but a) it might not be obvious what that means and b) is it worth it?
From: firstname.lastname@example.org (Dave Bakin);
Sent at Fri, 10 Apr 1998 13:01:21 -0700;
Re:"out of process servers for IIS"
Here are a couple of random points:
The article's point about scalability is that with MTS they've got this 'reusable pool of objects' and an object lifetime is carefully controlled: If you write an MTS aware object its actual bits don't exist between calls to it (no state). Thus there are only objects existing for those method calls that are currently in progress, not for all 'sessions to the page' that are currently in progress. If you're web server is serving 10000 users but is only doing the actual processing/computation for 10 at a given instant then you only have the bits for 10 objects around, not 10000. So that's what they mean by MTS scaling well where your arbitrary OLE server might not. Whether this is an issue or not probably depends not so much on Frontier, but on the usage of Frontier for a particular site or page.
A second point is that IIS is careful about security, and if you have a site that is requiring logon authentication, then I think IIS can -- if configured, or registry entries set for the OLE server, or something (I'm not sure) -- enforce security by making sure that it launches an OLE server in a new process with that login instead of using an existing process with a different security context. But I think the way this works is that the OLE server has the opportunity, in the registry or some other way, to specify the kind of security constraints it wants the system to enforce, therefore this is optional (assuming you're set up in the right way).
There might be other OLE server issues that aren't obvious except to a COM consultant - someone who has actually deployed these servers in production. But that isn't me. Sorry!
Jonathon Peterson writes: "The Internet is built on nothing but freeware. That's the power of grad students looking to do something cool."
From: email@example.com (Steven C. Den Beste);
Sent at Fri, 10 Apr 1998 12:46:26 -0700;
The Java Balloon
The Internet is not even close to being run exclusively by freeware; all of the hardware which moves packets from place to place in all those big expensive fiber-optic networks, and all of the software which takes care of that hardware, is proprietary.
There's nothing wrong with any of this, of course, it just proves that the lines between open and closed source aren't very clear.
From: firstname.lastname@example.org> (Jonathan Peterson);
Sent at Fri, 10 Apr 1998 11:48:48 -0700;
Re:The Java Balloon
Dave, the difference is one of empowerment. If I don't like the features the white Unix god puts in, I can hire other people to change it. If a new feature is critical to my biz, I can extend free software, otherwise I'm stuck waiting for my critical need to bubble to the top of the product future feature stack.
What could you have done with Frontier early on if you had the source code to the Finder and AppleScript? Frustrated that Adobe hasn't put Apple Event hooks into their apps? Don't be, build the hooks yourself (and now that there is a freeware image processing application, you can).
The Internet is built on nothing but freeware. That's the power of grad students looking to do something cool. Microsoft was going to take over the net with proprietary MSN, remember? They couldn't possibly keep up with the pace of development on an open system, and now they're an ISP.
But but but... We're just like O'Reilly. We employ people who work on some code that's publicly distributed in source form, and we also employ people who work on code that is not. What's the difference?
Further, Frontier came before AppleScript, so it would have been impossible for us to have its source code early-on with Frontier.
Further, what if we added the features to the Finder and Apple didn't distribute the changes? Your argument completely overlooks this essential aspect of human nature. People don't want to play experiments with their computers, at least not most people. DW
Just a note: the acronym "RPC" is reminiscent of obsolete, never-really-worked technology, in my world.
From: bryce@Colorado.EDU (Bryce);
Sent at Fri, 10 Apr 1998 11:48:33 -0700;
Re:XML Becomes Invisible
Now I personally am not so shallow as to notice acronyms, but if you want to attract the large contingent of developers of all shapes and sizes and ages who go for sexy tech, you might want to come up with a sexy name.
I dunno-- maybe such a cheap marketing ploy is beneath you.
I don't know if you've seen this yet today.
From: email@example.com (Raines Cohen);
Sent at Fri, 10 Apr 1998 11:54:10 -0700;
On one hand, out-of-process is good because it protects the web server process; see:
From: firstname.lastname@example.org (Dean Chouinard);
Sent at Fri, 10 Apr 1998 11:08:51 -0400;
Out-of-process Components in ASP
On the other, it has more overhead.
I think the focus of this article is narrow. Sure, in-process components require less overhead, but what if I want to use the services of Frontier or any other Microsoft application -- Inter-application communication has to be out-of-process.
What if I want to run Frontier on one server, IIS on another, and use DCOM to communicate between the two. How will that scale.
I'm not sure whats the difference between out-of-process application and out-of-process component. Frontier is an application.
These questions are rhetorical, I'm still learning about this.
I like the Frontier COM server. By implementing COM, you have magnified the usefulness of Frontier in the Windows environment. Before COM Frontier was an island in my system; now it is part of the community.
I don't think the situation is nearly as bad as you appear to think.
From: email@example.com (Michael Winser);
Sent at Fri, 10 Apr 1998 11:10:14 -0400;
Out of process servers for IIS
First of all, the typical out of process application is something like Word or Excel. These apps were obviously not designed with server usage in mind. They tend to be fairly slow to start up. Imagine firing up a copy of Word for every page hit!
Second, it is worth understanding that the object loaded by IIS can be inproc and still talk to another process to do the actual work. This is how almost all database access is done. SQL server runs as a service on NT and an inproc object (ADO) is used to do database transactions.
Finally, MTS (Microsoft Transaction Server) is a key part of building robust web applications, especially for commercial use. MTS can only work with inproc objects. However, as I mentioned above, these objects can do their work by talking to a server application. If they can implement transacted operations on the server app then they can probably play well with MTS.
Imagine the following. A simple inproc com object that is used to talk to an instance of Frontier running as a service (on the same machine or even a different machine to spread the load). This com object exposes methods that can be called from ASP pages. These methods then use some RPC to actually do the work on the other end. The complexity and efficiency of this solution depends quite a bit on the multi-threaded-ness (what a word) of Frontier as a server app.
Caveat scriptor: I am not an IIS or MTS expert.
|This page was last built on Friday, April 10, 1998 at 6:46:13 PM, with Frontier version 5.0.1. Mail to: firstname.lastname@example.org. © copyright 1997-98 UserLand Software.|