Damien Katz

Sounds too good to be true...
Keep coding man, I can't wait!
- T

That's very well written and very suscinct. It's good that you end with practical sample applications that Couch would likely support well. Other than that, even though Notes wasn't built for the web from the ground up, it was extended to the web (and before a widely adopted open standard like php came), so is the Couch paradigm going to be that much better?

Sounds cool, but isn't this very much like Zope? Maybe Zope is a good starting point, lets add offline and a good IDE (I think Zope supports WebDAV that might be useble as well) to it and we're good to go.

Hi Damien

It sounds great!

But i'm a little surprise that your project has turned into a scaled down version (or a reinvention) of Domino. Besides the great challenge for you ..what does Couch bring ..that is not Domino in a simular way?

brgds
Jesper Kiaer
http://www.jezzper.com

PS Don't get wrong ..I am very impressed of what you are doing! :-) (again)

This look really cool Damien!

I'll be very interested in seeing where this goes! Please keep us apprised of it's status and maybe even give us some screenshots and demos and such so we can contribute!

Do you have a financial backer? How do you plan to sel this (great) idea to developers who have open source products that do a lot of what this does? (albiet not as good...) Can you describe the UI? Will you be using an existing platform? (Eclipse, NetBeans, XULRunner?)

As much as I love python, please don't use Zope. Zope 3 is a huge improvement but zodb likje most odbms is tightly coupled to python.

I second xulrunner as the local platform and would love an xpcom binding for couch. This would expose couch to javascript in mozilla. Since mozilla 1.9 will support python for xull apps (not web apps), I'll settle for python bindings.

One thing I miss in Notes that was in several flat databases bofre Notes is a previous row reference in view columns. A parent ref would be nice too.

David - How will this be better than Notes/Domino? Well, it won't be burdened with 20 years of cruft that comes from being designed as a fat client. I'm trying to throw out the old Notes stuff that doesn't make sense, and keep the things makes Notes and Domino powerful at its core.

Gert - I've looked at Zope in the past, and it misses the mark for web development IMO. Zope uses a OODB, meaning its a DB is meant to provide a seamless persistance layer for Python. CouchDb is different, its meant to be a document database with robust reporting capabilities. I think this is a better match for the web.

Jesper - Yes, its very similar to in concept to Domino. I'd consider it to be Domino done right.

Jake - No financial backers and I don't have any business plans for it. Yet. But I do know one thing people will pay for: Offline. I can probably give everything else away and just charge for the offline component. But like I said, I have no plans yet. I just want to build this thing.

IMHO, If you build a db repository that one can embed that handles online & offline access smoothly, that would be marketable. Sorta like the MySQL of replicated datastores... Charge for the ancillary services. Everything else would be gravy :-)

Damien, sounds like the lightweight nature of this and the focus on offline capability makes it a perfect candidate for mobile porting... given that CRM is one of the areas you're foreseeing this being useful for, remote and offline access via PDA might be something to specifically account for in the framework design; this is an area that Domino often gets slammed on, since this type of functionality wasn't even on the horizon until long after Domino had already become bulky.

I like the idea of building your IDE as a web app hosted your own app server. I love self-referential systems like that (C compilers written in C, etc.).

I think it also makes a lot of sense to have one UI experience. Using a browser even for off-line use might seem counter-intuitive, but the two client schism is still a problem for Notes and Domino. Writing an application for both Notes and the Web can be a real headache.

The browser has a lot of shortcomings, but now that Ajax "best practices" have matured, you can probably get a lot of mileage out of the browser for both your IDE and the end-user experience. I am a big fan of rich clients, but in this case, you are probably right to introduce a rich client as late as possible. When you do, you should consider Eclipse RCP. It's a great foundation for an IDE.

Nice job Damien. I think the idea of a browser based IDE is interesting. I recently was looking at a project called Wheat that has a great tag line:

Every object has a URI, the virtual machine is a web server, and the development environment is a Wiki!

The idea of a collaborative development environment that could be used either online or offline using a Couch based Wiki sounds like a cool idea to me.

wow! i've been following this blog for a while, since you were talking about Couch as some sort of network / server infrastructure that sounded kinda like Google's back end.

then i noticed you talking about some kind of formula language.

then, pow! very interesting. i'd love to beta test when you get around to that.

i already have an application prototyped in Notes that i have been thinking about porting to another platform, but it needs offline capabilities desperately. i was even thinking about selling it (i've been back and forth whether to open source it).

when you get far enough along let me know, i'd be more than happy to try to develop the same app for Couch. no better way to test your dev tools, i think, than to have ppl start to develop for it. ping me, i'll fill you in if you are interested.

You mean to tell me that I could build my applications in Couch and not spend the majority of my time undoing/hacking around what it's trying to do *for* me so it can tell my v6.5 customers that the application would still be compliant with v4.6?! ;-)

First time/long time, and I (and I'm certain that I echo the majority here) can't wait to get crackin'!

It is a little bit funny that your search for what Couch should be ...ended back at something Domino like :-)

I guess there is an attraction..a beauty in what Notes/Domino is. Not the actual implementation but the idea that everything is built around a note/document etc.

I have been looking for a simple Java web framework but no one seems to get near the ideas behind Domino.

I like the Notes/Domino concept so much that I actually started building a Java web framework myself built around some of the ideas from it. I want it to be really really simple so I actually chose the name SimpleD for it :-)

If it ever gets finished ..that is another story.
(kids, work..you know..)

brgds
Jesper Kiaer

http://www.jezzper.com

Can we presume that part of the cruft you're throwing out is the summary buffer limit?

Tim - A lot of the architecture and design decisions I'm making are with portability to other platforms in mind. And lightweight devices are definitely included.

Dave - Initially I want a very simplified, end user centric IDE. This way I don't compete with mature, powerful IDEs target at professional software developers (that's a very crowded market). Instead I'm competing with IDEs that let end users (office manager, secretaries, etc) build simple, replicatable Web apps, and I think the browsers can meet that requirement. And yes, I hope to have it self hosted, where the IDE is just another Couch app. That would be cool!

Richard - CouchDb has no limit on per document summary data.

"I'd consider it to be Domino done right."

Man, you must not have gone out of the house since you started working on this...

..otherwise, your ego would've never got back through the doorway....

Shame on you

Yes, my ego is big. But maybe not as big as you think.

Firstly, I'm not talking about Notes & Domino as a whole, there is no way in hell I can rebuild that. Actually there is no way in hell I can even build all this. I'm going to reuse lots of infrastructure and I'm going to need lots of development help.

When I say "Domino done right", I am specifically talking about Domino HTTP and ease of building Domino Web applications. The whole thing, while brilliantly executed, was flawed in its quest to always translate Notes applications for the Web, rather than focusing on being a first class web application platform on its own.

And being grafted onto an already mature code base and carrying forward functionality intended for a propriatary fat client, it carried along a lot of baggage from decisions made long ago. I've developed many Notes applications for the web. This much I know: It doesn't have to be this difficult.

I think if IBM would have placed more priority on making it a first class web development platform separate from Notes, it would be a major web server player today. But for some reason that didn't happen.

So I'm building Couch because I believe the model that Notes pioneered is an unappreciated one, one that hasn't been explored enough. I'm building what I know works (i.e. I'm not very imaginative). And yes, I think I can improve on an already great design. I'm egotistical that way.

Cool idea.
Some thoughts:
- Are you planning to have a import/export/sync facility to POD (plain old databases)?
- what's about a @formula converter?
- How can we help?

:-) stw

I love your emphasis on end-user authorable design. I've always thought Lotus's R4 move away from this (not by the introduction of their designer client for advanced stuff, but by the removal of simple design from the end user license) betrayed a fundamental lack of understanding of why Notes kicked ass.

The days of unlocked R3 environments before admins got "smart" and stopped allowing anyone to put anything on the server was a magical time. If you can't prevent bad programming (e.g. @Today in selection formulas) from degrading the server, you'll only encourage IT depts to be lawnmowers of any grassroots support you'd hope to enjoy.

Can't argue with Notes' great success, and the web's, but putting them into the thin client vs. bloated client model, and the goals of Couch, it makes me wonder whatever happened to Lotus eSuite? Maybe not exactly the same but similar high level goals, would you say? Back around 1998 I thought having a spreadsheet on the web for same intended users as yours would have been big, but fizzled out and was discontinued. But back then, "thin client" meant "network computer"- not necessarily mobile phone, or pda. No one actually wound up buying any network computers- the market chose to stick with fat clients instead. I would question your viewing Notes/Domino as too fat- something that might be irrelevant, as far as the market's concerned (not that any of us know the Couch market yet, granted- might be a ton of people out there waiting for this or some highly anticipated app that Couch does the best way). Notes is cool, as you said, by not being relational yet still valuable at the same time; but if you're going to reinvent that wheel, unappreciated as it is, there must be some killer apps that take advantage of Couch's thinness that Domino isn't agile enough to pull off, and where people actually demand thinness in that case.

We are all conditioned by our past, so are all compelled to compare what you are building with what we have known and believed in as being "the" solution to putting applications on the web. Ray Ozzie reinvented Notes with Groove (and took groupware back to being a small group application but interestingly broke us out of Internet=browser mindset). Andrew Ressler reinvented Notes brilliantly with Radnet WebShare (smartest person I have ever met). It was sad that Radnet didn't survive. (My suggestion that WebShare should have an NSF interface didn't go over so well, but that probably wouldn't have mattered anyway.) And now Couch. I think all three approaches honor what drew us all to Notes in the first place, the ability to quickly build applications that people can use from anywhere. The mechanics of how that is done is what distinguishes the four efforts.

Stephan:

I have no plans yet for an importer/exporter/sync thingy other than to export and import databases and documents as XML. But I like the way you think, and I definitely see such things being built, but there are too many unknowns at this point to know what such a thing would look like.

An @Formula converter is a terrific idea, but like the import/export facility its a bit premature right now.

I've been getting lots of offers for help. It's fantastic! However, right now the help I most need is input and feedback. I've already gotten some good feedback about Fabric that is prompting me to make changes in the syntax. Soon enough though, I'll be ready for all kinds of help: development, testing, documentation, setting a project website Wiki, etc. Lots to do!

John:
I think you "get it". I think many Notes professionals, particularly those around in the R3 days, understand something that IBM doesn't seem to get: Notes got popular because it was so damn easy.

I started off my career doing Notes consulting. Almost every company I did work for had numerous Notes applications that were developed by secretaries, office managers and sales people. And some of the applications became core parts of the companies' businesses. And they kept getting modified and tweaked as the business needs changed.

This is how Notes infiltrated so many companies: people solving their own problems and building just what they need, when they need it. Truly software that evolves with you. And once it's in place, the custom applications seem to mulitply, and then it's nearly impossible to get rid of Notes. :)

But the funny thing is, I heard very little about this phenomenon at Iris/IBM. These "amateurs" building little in-house applications to run their companies, they pretty much weren't on the radar screen at all. The people who were working on the Designer client certainly were aware, but on the Client team (which the Designer was part of) *everything* seemed to revolve around mail, and the Designer effort was seriously underfunded and developer and QA resources were constantly pulled off to work on other things (usually mail).

But, hey IBM managed to sell 200 trillion licenses anyway, so maybe they made the right decision. But they still dropped the ball on the end user development aspects, and there is still lots of unexploited opportunity there.

David & Kent:
Your thoughts mirror many of my own.

Couch won't replace Notes, and therefore also won't replace Domino. It also won't replace other web development tools, relational or otherwise. But that's ok, for Couch to be successful, its not necessary for others to lose.

T-shirt idea:
"Couch sayeth: Schema shmeema!"

err, Sorry. Made _me_ giggle, anyway. ;-)

Damien, thanks for the clarification.
I, of course, understood what you meant and the context in which you placed the quote.
I was just...um...well.. just me being me.

The only thing that still rankles about the whole Domino/Notes thing is that somehow Domino got branded as the application development platform.

Notes was always and ever about application development. IT WAS NEVER ABOUT SOME FAT--SS CLIENT THAT YOU COULDN"T TAKE HOME TO HAVE YOUR IT DEVELOPMENT MANAGER MEET AND EVEN NEVER MORE WAS IT ABOUT E-MAIL!!!

Ok, sorry. Time to take my meds.

I miss you man. Best of luck with your Couch stuff...

P.S. Have you developed a jump shot yet?

Damien,
put me on the helpers list.
Have you ever come across Thingamy? My gut feeling says Fabric, Couch and Thingamy would be a terrific strong brew.
:-) stw