Couch Book Progress
It's been a few weeks since we released the initial chapters of O'Reilly's CouchDB book. The next release ought to be happening "soon" - There's more work to be done on it, but the major skeleton of the chapters has been written, and the figures drawn.
I'm posting the outline of the next section, so that people's expectations about the book are inline with what we provide. The next released section will be a few chapters walking through the creation of a simple web app (yes, a weblog). The point is not to teach Ajax, but to teach the CouchDB HTTP API in a hands on way.
Yes, all the code runs, and the in-progress version of the example blog is even available on github. One of my main tasks between now and the release is to make the installation proceedure much easier. So you are welcome to play now, but if it seems messy or confusing, please try again when the chapters are released.
If anyone has input on pacing and ordering of the example
application I'd love to hear it. I think this section should be
relatively fast-paced, and focussed on how easy it is to build
applications using CouchDB, with the occasional bits of API
specification or deeper material presented as asides and sidebars.
h3. 2. Learning CouchDB
h4. Touring the Futon Administration Interface
A survey of CouchDB's http api, through the eyes of Futon. Touches
on most of the core concepts. This material is might be better suited
for Section 1, as it does stand alone as an introduction to the http
API.
h4. Introducing the Example Blog
Why we chose JavaScript for the book examples. Also: even if you are
storing scientific data, what you'll learn by following along is
applicable (views, document semantics, design docs).
h4. Setting Up Your Development Environment
Getting the example app to your local drive, and being able to push
it to CouchDB. Also cover design-doc / application correspondence
while explaining the couchapp script.
h4. Storing Blog Posts in CouchDB
The JSON doc format, validation functions, plus the HTML and
Javascript it takes to support the blog post form. (This will be a
longer one, but the code for it shouldn't be changing.)
h4. Rendering Documents with CouchDB Forms
What forms are. Why they have the single-doc constraint. Mention
that you can write raw JS, but then show CouchApp's require / include
system as a best practice, using templates and a basic templating
engine.
h4. Browsing Recent Posts with Map Views
Introduce a simple view by published at date. Show basics of
pagination. Explain the view-query API by looking at the JavaScript
client API used to query the view.
Mention that there will be forms for views in the future, so that
this page can be served as HTML as well. (Forms for views should be in
Couch 1.0 - and thus in the book, so polishing this section now is
lower priority. Pagination will also be easier with view forms)
h4. Accepting Reader Comments
The JSON format, validations, and the HTML form for updating a
comment. This is a short one, mostly repeats what happened in "Storing
Blog Posts in CouchDB" but its the second time, so we can touch on
different aspects and explain other API details. (One neat difference
is that we're adding Javascript behavior to a page that was rendered
as a CouchDB form, rather than a static document.)
h4. Listing Related Comments Using View Collation and Compound Keys
(I didn't say "joins") ;)
Currently this uses an Ajax query that lists comments by [post_id,
date]. When view forms are available, we'll change it to "Christopher Lenz's
query,":http://www.cmlenz.net/archives/2007/10/couchdb-joins which includes the post itself, so the whole thing can be
rendered in a single request. This chapter works on the same files as
the previous chapter, they could be "joined" but I think this is
material we should take the space to cover as clearly as possible.
---
If you've gotten this far, then you must be seriously interested in CouchDB. If you are, then you should familiarize yourself with the test suite. It's written in JavaScript, so anyone can understand it, and it's the main gateway for new features. Plus, code speaks louder than words.
by Charlie, 2009/01/05 23:51:20 +0000
Very Cool - can't wait to read it!
We are overdue for this kind of software.
As an aside, I had no luck attempting to run CouchDB on my mac using macports. Then I think I came across something in a post somewhere saying that the current port doesn't work. Then I tried downloading CouchDBX (the 0.8.1 version). That worked ok but hung when attempting to clone the demo twitter client (ever time I tried it stopped after 133 documents, and never got any design docs).
I haven't yet ventured into building from scratch, which I suppose would be the next thing to try.
I guess what I mean to say is, getting it to install and work right away would be awesome :)
by J Chris Anderson, 2009/01/06 03:20:27 +0000
@Charlie "CouchDB svn trunk":http://svn.apache.org/repos/asf/couchdb/trunk/ is definitely the way to go currently.
Yes I do think these sorts of apps have been a long time coming. It's too bad Dashboard widgets didn't really make it (although they have a second coming as iPhone apps).
I hope you get your installation issues settled. We're working on an 0.9 release, wish I could give you an exact date, but suffice to say there has been a lot of work done since the last release.
by Charlie, 2009/01/07 00:53:13 +0000
@JChris I'll definitely get there soon. Part of the delay stems from a fear of getting into installing a new thing when I don't know for sure where I should put it in the filesystem, how exactly to set it up (as a daemon on startup? with its own nonroot user? Is there maintenance to do like log rotation etc...)
Basically I'm not a very good sysadmin over my own macbook - I don't have much of a system going so things like this feel like a bit of a project. I should really find some kind of remedial crash course on that sort of thing...
It does seem a bit ironic that something like couchdb is in an SVN repository rather than something like mercurial or git - my first reaction once I got the cloning piece of the puzzle was "oh, it's like the git of databases!"