Resources In Various Frames of JSON

I've been mulling over the role(s) JSON should play in representing RDF the last couple of days (well, the last year or so really).

Having worked with RDF for some years now, more or less full time, in specific contexts (legal information, and lately educational data), I'm getting a hang of some usage patterns. I'm also following the general Linked Data community (as well as various REST endeavors, Atom applications, and various, mostly dynamic language based, platforms). In all of this, I've made some observations:

1. If I want to use RDF fully, anything but a full RDF API (Graph navigation, Resource and Literal objects and all) often cause a lot of friction. That is, unless I already know what "framing" I need of the data, I really need a whole data set to dig through, for purposes of e.g. finding incoming relations, looking up rdfs:label of types and properties, filtering on languages, irregular data (such as dc:creator used both with literals or URI references) and so on.

2. Encountering general JSON data (not RDF specific) on the web, I sometimes come across quite crude stuff, mostly representing a database slice, or gratuitous exports from some form of O/R mapper. It may look meaningful, but often shows signs of ad-hoc design, unstable modeling without "proper" domain understanding and/or implementation leakage. However, the data is accessible for most web programmers without the need to get the domain properly, no matter how poor this data representation may be. JSON is language native. The use case is to have users (programmers) be able to spin around a specific, boxed and digested slice of data. Ideally you should also be able to find and follow links in it. (Basically the values which match something like /^(http(s)?:|\/|\.\/)\S+/ ...).

3. If I know and control the data, I can frame it in a usage scenario (such as for rendering navigable web pages with summaries of entities) based on a specific domain (such as an article document, its revisions and author, etc.). Here is great potential for reducing the full data into a something, raw, more (web) programming language native. This is where a JSONic approach fits the bill. Examples of how such data can look includes the JSON data of e.g. New York Times, LCSH. The Linked Data API JSON is especially worth mentioning, since they explicitly reduce the data for this casual use so many need.

Point 1 is just a basic observation: for general processing, RDF is best used (produced and consumed) as RDF, and nothing else. It can represent a domain in high fidelity, and merges and is navigable in a way no other data model I've seen supports.

Point 2 is about quick and sometimes dirty. Cutting some corners to get from A to B without stopping for directions. You cannot do much more than that though, and in some cases, "B" might not be where you want to go. But it works, and if the use case is well understood, anything more will be considered waste for anyone not wanting the bigger contexts.

Point 3 then, is about how to go from 1 into 2. This is what I firmly believe the focus of RDF as JSON should be. And since 2 is many things, there may be no general answer. But there is at least one: how to represent a linked resource on the web, for which RDF exists, as a concise bounded description, showing inherent properties, outgoing links and per application considered relevant incoming links. And how to do this in JSON in high fidelity but immediately consumable by someone not wanting more than "just the data".

Many people have expressed opinions about these things of course. You should read posts by e.g. Leigh Dodds and Nathan Rixham, and look at some JSON Serialization Examples. Also monitor e.g. the Linked JSON W3C mailing list and of course the ongoing work of JSON-LD. Related to the Linked Data API and its "instrumental" JSON is also a recent presentation by Jeni Tennison: Data All the Way Down. It's short and very insightful. End-users have different needs than re-users!

Early on (over a year ago) I drafted Gluon. I have not used that much since. A related invention I have used though, is SparqlTree. While it isn't really a mechanism for defining how to map RDF terms to JSON (but to formulate SPARQL selects digestible into compact results), it does so quite well for specific scenarios. It is very useful to create frames to work on, where code paths are fully deterministic, and where there is a one-way direction of relations (which is needed in JSON trees, as opposed to RDF graphs where we can follow rel and rev alike). Admittedly I've done less than I should to market SparqlTree. But then again, it is a very simple and instrumental solution over an existing technology. I recently gave a glimpse of how I use it in a mail concerning the "Construct Where of SPARQL 1.1" .

Reflecting on all of this, I'm quite convinced that anything like RDF/XML or Turtle is beyond what JSON should ever be used for. That is, support for all kinds of general RDF, using prefixes (whom I love when I need to say anything about anything) and exposing the full, rich, internationalized and richly and extensibly datatyped world of literals is beyond the scenarios where JSON is useful. If you need full RDF, use Turtle. Seriously. It's the best! It's rather enjoyable to write, and I can consume it with any RDF API or SPARQL.

The only case I can think of where "full RDF in JSON" might apply is for machine-to-machine data where for some reason only JSON is viable. For this, I can see the value of having Talis' RDF/JSON standardized. It is used in the wild. It is reminiscent of the SPARQL results in JSON, which for me is also quite machine-like (and the very reason for me inventing SparqlTree in the first place!). I'd never hand-author it or prefer to work on it undigested. But that's ok. If handed to me I'd read it into a graph as quickly as possible, and that'd be dead simple to do.

So where does this leave us? Well, the Gluon I designed contains a general indecision, the split into a raw and a compact form. The problem is that they are overlapping. You can fold in parts of data into compact form. This is complex, confusing and practically useless. Also, the raw form is just another one in the plethora or more or less "turtle in JSON" designs which cropped up in the last years. I doubt that any such hybrid is usable: either you know RDF and should use Turtle, or you don't and you want simple JSON, without the richness of inlined RDF details.

My current intent is to remove the raw form entirely, and design the profile mechanism so that it is "air tight". I also want to make it as compact as possible, true to RDF idioms but still "just JSON". A goal will still also be that if present, a profile should be possible to use to get RDF from the JSON. This way, there is a possibility of adding the richer context and integratability of RDF to certain forms of well designed JSON. This of course implies that Gluon-profile compatible JSON will be considered well designed. But that is a goal. It has to look good for someone not knowing RDF!

I have a strawman of a "next generation gluon profile" in the works. I doubt that you can glimpse my design from that alone, but anyway.

Some things to note:

• The 'default' feature will be more aligned with the @vocab mechanism of RDFa 1.1 (and JSON-LD)
• Keywords ('reserved') can be redefined. There are preset top-level keys, but that's it. (A parser could parameterize that too of course.)
• No CURIEs - every token is "imported" from a vocabulary.
• Types will be powerful. They'll determine default 'vocab' for a resource description (i.e. JSON object), and you can also import terms locally for a type (so that a Person title is foaf:title although 'title' is globally from 'dc').
• If there are multiple values for a term (i.e. multiple triples with the same subject and predicate), a defined prefix or suffix will be added to the term. This is an experiment to make this nagging problem both explicit and automatic.
• The 'define' will be reduced to a much less needed component. Using 'autocoerce', pattern matching on values will be bravely used to coerce mainly date, dateTime and URI references to their proper types.
• Incoming links can be represented as 'inverseOf' attributes, thus making it possible to frame more of a graph as a tree.
• Named bnodes are out (though they might be snuck in via a "_:" link protocol..). Anonymous bnodes are just fine.

This is a design sketch though. Next steps are to work on adapting my test implementations and usage thereof.

An auxiliary but very interesting goal is the possibility of using these profiles in a high-level API wrapper around an RDF graph, making access to it look similar to using Gluon JSON as is (but with the added bonus of "reaching down" the abstraction to get at the details when needed). (This is the direction I've had in mind for any development of my nowadays aged Oort Python O/R mapper. More importantly, the current W3C RDF/Structured Data API design work also leans towards such features, with the Projection interface.)

(Note that profiles will reasonably not be anything like full "JSON schemas". It's about mapping terms to URI:s and as little else as possible to handle datatyping and the mismatch between graphs and JSON trees. There is a need for determining if a term has one or many values, but as noted I'm working on making that as automatic as possible. Casting datatypes is also needed in come cases but should be kept to a minimum.)

Finally, I really want to stress that I want to support the progress of JSON-LD! I really hope for an outcome to be a unification of all these efforts. The current jungle of slightly incompatible "RDF as JSON"s sketches is quite confusing (and I know, Gluon is one of the trees in that jungle). I believe JSON-LD and the corresponding W3C list is where the action is. Since there is work in JSON-LD on profiles/contexts, and a general discussion of what the use cases are, I hope that this post and my future Gluon profile work can help in the progress of this! For me Gluon is the journey and I hope JSON-LD is the destination. But there are many wills at work here, so let's see how it all evolves.

The Groovy Times

It has been so long since my last post here. I've twittered away like the rest of my peers. I guess I could dish out details from my personal life of the past year now. To examine my interrupt. I won't.

I've been using Groovy a lot in my work on the Swedish Legal Information System. While the things I find interesting in this work deserve many separate posts, I'll just spend this one to drop some nice stuff about groovy.

"Why Groovy", you might ask? Oh dear. For "political" reasons (this is an entire topic of its own), I have to use Java. But the language Java is often so much overwork and ceremony; riddled with convoluted ways to make explicit patterns and formalisms. Dynamic languages are pragmatic. Sure they have flaws, but I find the compromise acceptable. There are probably thousands of articles discussing this, and I prefer to debate it elsewhere (mostly with friends over lunch/dinner/beer).

I must deliver the bulk in Java, and Groovy is impressively close to Java, but with so much more expressive power (ease of use). (More than necessary? My pythonic side says "probably".) I've worked a lot with Jython the last decade, and some with JRuby. Groovy trumps them both (IMHO) when it comes to java library and java culture interoperability. I can spike, explore, and make tests in groovy. Then I add all this horrendous checked exception handling, spinkle some semicolons for the grumpy old compiler, and finally explode-a-pop all the def:s to bulky types when things need to "harden" (to fulfil the contract of delivering .java...; the tests are left in groovy). Not a big deal (especially since I use an editor that makes code editing a breeze). And groovy cross-compiles nicely with traditional cruft.

So what have I used more specifically? Spock. Check it out. Rewrite ten of your JUnit4 tests in it, and if you go back, write ten more. Don't go back. If you're already testing with say JRuby, I won't push you, but I assure you Spock is worth looking at. Specs become liberatingly clear and thin. Data-driving some of them is pure joy. Mocking is dead simple. (Sorry, I won't put code examples here now: look at the spock docs for that. Try them out!)

For building, we do not use Gradle (not yet at least), but that scary beast of mindnumbing declarativity (which I'm usually for), dreadful xml (which I can handle due to prolonged exposure) and conflated purposes (no excuse here) known as Maven 2. It seemed paramount in the surroundings when we started, and won't go away soon. I use it as little as possible (and it is quite impressive when you let it do its thing).

Which leads me to the last thing I want to mention: how to use Groovy's very convenient, builtin Grape system (and its @Grab mechanism) together with my local maven2 artifact repo. That one in <~/.m2>, where all my local packages have been mvn install:ed (along with the umpteen dependencies).

The thing was, when I started, I naively thought things would kind of work at least semi-automagically. I've spent my time in CLASSPATH hell. I wanted groovy to tap into the local m2. Then I was disillusioned again, and attempted to run experimental groovy scripts via GMaven. Didn't fit my use cases at all (that thing is great at compiling, I leave it at that). I shellscripted the path from mvn dependency:build-classpath, then built pathing jars, then just felt quite uneasy (such moves work, but it's not particularly clean).

When Grape appeared I tinkered with the Ivy config in <~/.groovy/grapeConfig.xml>. It surely looks so simple. I couldn't figure it out. Benhard could. Neat.. But alas, that solution copies all the dependencies from the local m2 repo to grape's ivy repo. And I could not get it to grab my new local SNAPSHOT-stuff as they landed either (in spite of eleventy ivy attributes claiming to force all kinds of checks).

Then it appeared, from a combination of fatigue and taking a step back (cue magic "aahh").

Use Groovy's Grape with your Local Maven File Repo

Locate $HOME/.groovy/grapeConfig.xml. If it doesn't exist, see the Grape docs for how the default version should look.

Then add, directly after (xpath) ivysettings/settings, the following directive:

  <caches useOrigin="true"/>

And, in (xpath) /ivysettings/resolvers/chain, after the first filesystem, add:

<filesystem name="local-maven2" m2compatible="true">
  <ivy pattern="${user.home}/.m2/repository/[organisation]/[module]/[revision]/[module]-[revision].pom"/>
  <artifact pattern="${user.home}/.m2/repository/[organisation]/[module]/[revision]/[artifact]-[revision](-[classifier]).[ext]"/>
</filesystem>

With that in place (pardon the line width), Grape (and thus @Grab) will happily use anything it finds in your local m2 file repo, without copying the jar:s. (It will still download other stuff to use to the default ~/.groovy/grapes/, which is fine by me.)

That's it for now. There are lot's of cool stuff with Groovy, if you're in a Java environment and want to ackowledge that without giving up modern power.

Three years ago I looked at Scala for the first time, and found it quite interesting. Then I got a bit spooked by the academic machinations of it. Lately that interest has been quite rekindled though, and the future will show what will come of that. I am very happy to have used Groovy so far though, and I would certainly recommend it. Scala may yet be for tomorrow, Groovy is for the Java user of right now.

(Of course, I recommend to continuously look beyond the JVM as well. Simplicity is hard to reach in increments without designing for it from the start. But things will reasonably evolve in most "camps".)

Labelled Reduction as A Good Thing

One small thing (of the many) in Python 2.6 I like (and have waited for since it appeared as a recipe), is collections.namedtuple. It is very useful in itself, but the fact that the stdlib has been adapted to use it throughout is quite nice. Consider the following code:

from urlparse import urlparse
print urlparse(
        "http://localhost:8080/doc/something;en?rev=1")

If run with Python 2.5, you get this tuple:

('http', 'localhost:8080', '/doc/something',
 'en', 'rev=1', '')

, whereas in 2.6 it is a namedtuple:

ParseResult(
        scheme='http', netloc='localhost:8080',
        path='/doc/something',
        params='en', query='rev=1', fragment='')

The last one is unpackable just like a regular tuple, but you can access the parts as attributes as well. This little "data struct" is quite handy, since I don't like to access tuples by index, but quite often pass them around and only access some piece of them at a time.

(With these you don't need to create full-fledged classes for every kind of instrumental data. (Sometimes coupling data and functionality in a single paradigm may be a coarse hammer, treating nails and iron chips alike..) Nor resort to the use of dictionaries where you really want a "restricted value lens", if you will.. But this is another rant altogether.)

Of course, there's lots more to enjoy in 2.6 (the enhanced property for decorator use, ABC:s, json, 2to3 etc).

On a related note, do check out Swaroop C H:s excellent and free books on Python (2.x + 3.0(!)): A Byte of Python. And if you're into Vim (you should be, IMHO) his new A Byte of Vim.

Resources, Manifests, Contexts

Just took a quick look at oEmbed (found from a context I was led to from Emil Stenström (an excellent Django promoter in my surroundings btw. Kudos.)).

While oEmbed is certainly quite neat, I very much agree with the criticism regarding the lack of RESTfulness, and that they have defined a new metadata carrier. I think oEmbed would work very well as an extension element in Atom Entry documents (who already have most of the properties oEmbed (re-)defines). Or by reusing (in such atom entry docs) e.g. Media RSS, as Stephen Weber suggested.

Granted, if (as I do hope) RESTfulness and Atom permeation on the web becomes much more well established (approaching ubiquity), this would be dead easy to define further down the line. (And signs of this adoption continue to pop up, even involving the gargantuans..)

But since it wasn't done right away, oEmbed is to some extent another part of the fragmented web data infrastructure — already in dire need of unification. It's not terrible of course, JSON is very effective — it's just too context-dependent and stripped to work for much more than end-user consumption in "vertical" scenarios. While oEmbed itself is such a scenario, it could very well piggy-back on a more reusable format and thus promote much wider data usability.

A mockup (with unsolicited URI minting in the spaces of others) based on the oEmbed quick example could look like:

<entry xmlns="http://www.w3.org/2005/Atom"
       xmlns:oembed="http://oembed.com/ns/2008/atom/">
  <id>tag:flickr.com,2008:/3123/2341623661_7c99f48bbf_m.jpg</id>
  <title>ZB8T0193</title>
  <summary></summary>
  <content src="http://farm4.static.flickr.com/3123/2341623661_7c99f48bbf_m.jpg"
           type="image/jpg"/>
  <oembed:photo version="1.0" width="240" height="160"/>
  <author>
    <name>Bees</name>
    <uri>http://www.flickr.com/photos/bees/</uri>
  </author>
  <source>
    <id>tag:flickr.com,2008:/feed</id>
    <author>
      <name>Flickr</name>
      <uri>http://www.flickr.com/</uri>
    </author>
  </source>
</entry>

The main point, which I have mentioned before, is that Atom Entries work extremely well as manifests of resources. This is something I hope the REST community will pick up in a large way. Atom feeds complement the RESTful infrastructure by defining a standard format for resource collections, and from that it seems quite natural to expose manifests of singular resources as well using the same format.

In case you're wondering: no, I still believe in RDF. It's just easier to sell uniformity one step at a time, and RDF is unfortunately still not well known in the instrumental service shops I've come in contact with (you know, the ones where integration projects pop up ever so often, mainly involves hard technology, and rarely if ever reuse domain knowledge properly). So I choose to support Atom adoption to increase resource orientation and uniformity — we can continue on to RDF if these principles continue to gain momentum (which they will, I'm sure).

Thus I also think we should keep defining the bridge(s) from Atom to RDF for the 3.0 web.. There are some sizzling activities on that respect which can be seen both in the Atom syntax mailing list and the semantic web list. My interest stems from what I currently do at work (and as a hobby it seems). Albeit this is from a very instrumental perspective — and as a complement, rather than an actual bridge.

In part, it's about making Atom entries from RDF, in order for simple RESTful consumers to be able to eat some specific Atom crumbs from the semantic cakes I'm most certainly keeping (the best thing since croutons, no doubt). These entries aren't complete mappings, only selected parts, semantically more coarse-grained and ambiguous. While ambiguity corrupts data (making integration a nightmare), it is used effectively in "lower-case sem-web" things such as tagging and JSON. (Admittedly I suppose it's ontologically and cognitively questionable whether it can ever be fully avoided though.)

We have proper RDF at the core, so this is about meeting "half way" with the gist of keeping things simple without loosing data quality in the process. To reduce and contextualize for common services — that is at the service level, not the resource level. (I called this "RA/SA decoupling" somewhere, for "Resource Application"/"Service Application". Ah well, this will all be clarified when I write down the COURT manifesto ("Crafting Organisation Using Resources over Time"). :D)

Hopefully, this Atom-from-RDF stuff will be reusable enough to be part of my Out of RDF Transmogrifier work. Which (in my private lab) has been expanded beyond Python, currently with a simple Javascript version of the core mapping method ("soonish" to be published). Upon that I'm aiming for a pure js-based "record editor", ported from an (py-)Oort-based prototype from last year. I also hope the service parts of my daytime work may become reusable and open-sourced as well in the coming months. Future will tell.

Representation Taxation

It seems my thoughts about Atom Entries will be delayed a while longer. This post begun life as a response to this "BlubML" post by Bill de hÓra, but quickly turned out to be about my view of data models, as used in the IT industry yesterday, today and tomorrow. And industry rather ridden with technology wars, which has reduced the status of information into mere fodder. Leaving meaning, shared and reused concepts, discoverability, integratability and hence interoperability in many ways just a far-off vision many never even have time time to think about (leaving us thrashing data in the technology trenches).

The mentioned post reflects (by quotation) upon markup and the perceived "angle bracket tax". I can definitely get that. But, as mentioned in a comment by Iain Buckingham, the issue at hand is probably mainly about what the markup is supposed to represent. (And now I totally leave the subject of lightweight text formats (which I like, by the way); I'll focus on data models, not syntax). Take the content model of XML. It is a mixture of plain text (useful for primitive data like numbers, dates and other "human language" constructs) and different structure blocks/markers (elements, attributes, pi:s, some more) who have no useful semantics apart from a label (sometimes in a namespace), ordering and nesting. This has been proven as very useful to represent documents, where documents range from books via web pages (also as application views) to vector graphics and so on.

The "defining nature" of documents is difficult to pin down, but it seems we get by fairly efficiently anyway. Mainly since this semi-structure is intended (and, incidentally, often indented) for one-way (sometimes layered) rendering of output in turn meant for human consumption, which actually works without information precision (albeit sometimes with more or less apparent negative concequences).

Documents aside, let's continue to data records. More or less the first use of XML was as a serialization of the RDF data model. This model has real semantics, where unique resources are described with statements, whose parts (subject, predicate, object) are either resources (all three) or primitive values (for objects). Unfortunately, RDF still struggles to emerge as a useful representation in the industry at large, partly due to its initial appearance as XML (from whose abundance of namespaces and literal URI:s even many XML neophytes has fled in panic), partly due to its data model being more academic (rooted in logics and AI) and thus less approachable than the (IMHO) less expressive but quite succinct object oriented, class based data model of most common modern programming languages.

This latter model has always been fragmented by language differences, and up close hard to pin down due to conflicting computer science details (OO encapsulation, coupling with function and message metaphors etc). And while objects often live in (connect as) graphs, their identity has been hidden, often being a mere memory address. Quite parallell with this, data has been pushed into the even more mechanical and implementation focused semantics of the relational model, in SQL databases, for decades. While these records do have explicit IDs (or composites, or...), they are for internal use within a given application context. Any usage beyond that must be secured by convention outside of that. The often cumbersome use of RDBs (where "splitting and slicing" all but the flattest records is needed) has been minimalistically remedied by the O/R-mapping tools in the OO languages, which has proven exceptionally useful in introspectable and/or dynamic languages. But it's mostly just a prettier surface upon the same old relational model. Signs of which can be seen by the specific restrictions (that many O/R:s have) placed upon the object semantics of the "host" language (inheritance, composition, coupling with functionality), and not the least the invention of SQL-like languages, mostly one per mapping implementation (although some credit should go to earlier efforts by the object database people for trying to standardize such things).

But there are very useful aspects of an OO-based representation. As a common denominator, JSON should be held up as an extremely useful format (albeit quite void of namespacing). It represents the "bare bones" data record format which is isomorphic with the OO languages (be it class- or prototype-based ones). And although a more narrow scope, it has more defined semantics than XML, since it explicitly differs between properties and lists (and does not introduce the artificial controversy of whether to use "attributes" or "elements"). But it's still a long way from the identifiable resources and datatyped or human language typed literals of RDF. Depending on context/domain needs, this may or may not be a painful point to realize (if not, JSON is indeed the simpler thing that actually works).

The attempt to express similar semantics in W3C:s XML Schema is to me a painful one, and in view of the complexities it adds and how little it brings in succinctness and clarity (and immediately/apparently useful general semantics) is a sign of that schema technology's failure. Use of RelaxNG isn't too bad though, and I find nothing inherently long in some formats, such as Atom, whose role is to bring a packaging and insulating channel format for resources (in the URI, REST, and most imporantly the RDF sense of the word). Atom defines a model which is scoped and simple (as in easy to grasp, limited in expressiveness). This is done with a text specification (RFC 4287 and its kin) coupled with a (non-normative) RelaxNG (compact) schema for easier machine use. For some things this can be a fitting approach. I don't believe its the best way to fully express records though (but perhaps to glean a narrow but useful aspect of such).

Now, let's turn the focus back to the objects often used as records with e.g. O/R technologies. It's been mainly through the emergence of adherence to the principles of REST, in turn (in practise) dependent upon URI:s and ideals of "cool URI:s" (for e.g. sustainable connectedness), that these objects have gained identities usable outside of a given, often ad-hoc defined, changing and quite internal (think "data silo") management of the data at hand. This process is the evolution of the Web as a data platform (which I believe should be the basis of a stable distributed computing platform, where one is needed), and it seems to me that the use of RDF is now a very promising next step. Since it embraces (indeed builds upon) this uniform and global identification scheme. Since it is "data first" oriented — you can use it to state things about resources which are meaningful even for machine processing without any invention/specification on your part. And since (due to URI:s and the semantic precision) it's less vulnerable to coupling with internal solutions, which "bare" O/R-mapped objects often are (in the same way, albeit less humongous, as automatically generated XML-schema defined API-fragments that hide in SOAP envelopes), due to implementation details (and sometimes a hard-wired and thus fragile URI minting mechanism).

With JSON, RDF shares the decoupling of types/classes ("frames" in AI) and properties of resources. But in JSON properties are simple, totally local keys, and not first-class resources themselves possible to describe in a uniform manner.  And JSON has other upper limits (such as the literal language mechanism), which can't be surpassed without adding semantics, "somehow" (e.g. conventions). This may work for specific needs (and very well), but is hard to scale to the levels of interoperability needed for a proper web-based data record format. (Similar problems riddles it's ragged step-cousin microformats, which also lacks formal definitions for how to handle (especially decentrally so).)

Back in "just XML"-land, as said, it seems to work for the semi-structured logical enigmas of XML that the above mentioned "documents" are, but as for data records, the XML format will be a serialization of some very specific, elsewhere explicitly defined model (document formats as well of course, but these often make some use of the more exoteric mixed content model XML supports). One such model is the "web resources over time" that Atom works for (coupled with the HTTP infrastructure and RESTful principles). Granted, such XML-based formats can be used without worrying too much about the mysterious XML infoset model (including the RDF/XML format, its free-form character aside). But to be useful, they need to be standardized and extensively supported by the larger community. Otherwise, apart from needing to design and implement the deserialized model from scratch, the creeping complexities that an ill-thought XML model design lets in (or forever locks out any extensibility from) may emerge whenever you want to express something idiomatic to your data.

That's all. This was mostly letting out some thoughts that have been building up pressure in my mind lately. And became yet another prelude to my thoughts still simmering regarding how to effectively use Atom with payloads of RDF. And also about some of the more crude ways, if you need to gain ground quickly (but more or less imperfectly). You can do that with JSON (fairly ok, but with JSON you're quite localized anyway, so why take the Atom highway if you only need to use your bike?), microformats (not ok in by my standards (again: where's your model?)), RDFa (if you have overlapping needs it can be brilliant), or with simplistic Atom extensions (could be very XML-taxating and thus risky; but done "right" it's kind of a "poor mans RDF", a bleak but somewhat useful shadow of the real thing).

(Admittedly this post also turned out to be a rant with questionable focus, but why not. Pardon my posting in quite a state of fatigue.)

Memes and Principles, Intent

This is a prelude to an upcoming post where I intend to speak about Atom Entries as some kind of "simplest thing that could possibly work" (for the specific purpose of representing manifests of resources ("resource" as the R in URI (and in RDF, of course) — i.e. the resources of the web, and the mind (perhaps even "the world", but I doubt it))).

The prelude is just some thoughts about a couple of principles.

First: the simplest thing that could possibly work. A phrase often quoted and often misapplied. This is common knowledge, and many clarify things by interpreting what "possibly work" means. I won't. I'd just like to rephrase it as "the simpler thing that actually works". It's useful since it both hints that there can be more complex/complicated things (see #3 and #4 in The Zen of Python) that don't actually work (for some arbitrary meaning of "work", admittedly), and (obviously) that there are simpler things that don't work at all (less or more simple than the one that works — this is the heart of the problem). I feel this phrasing avoids the confusion the original one often causes (it just seemed to be a simpler way of saying it that still works..). So, to repeat: just say "the simpler thing that actually works". It works.

Second: just a recap of an old joke of mine regarding the principle of DRY. I very much do prefer if programming practise adheres to that one. There's too much code that suffers from MOIST. Still, I advice you to avoid to DRY until SHRIVELLED. (Those sure are acronyms. Please do hover.) This is related to the first principle above, and I mainly mean that some practises, such as heavy use of metaprogramming, simply performs too much reduction to make the intent clear even to the appropriately trained eye. Mileage will vary, but as many others have already advised: don't be too clever. It will bite others, and you as well. (Ok; honestly, my main intent was probably just to make a pun out of a perfectly sane advice.)

I will probably repeat this.

[Side note: It's funny. Just this last week I've noticed I habitually misspell "intent" as "indent". I had to correct myself about three times while writing this. Obviously my prolonged use of Python has caused these two distinct words to conflate in my mind. I never indented that.]

Getting lxml 2 into my gnarly old Tiger; going on ROA

In case anyone might google for something like:

lxml "Symbol not found" _xmlSchematronNewParserCtxt

, I'd like to reiterate the steps I just took to get lxml 2 (2.1 beta 1 to be precise) up and running on OS X 10.4 (which I haven't yet purged in favour of the shiny Leopard for reasons of fear, uncertainty and doubt. And while pleasurable, quite mixed impressions from that cat when running my wicked new iMac (and my funky old iBook, in case you're dying to know)).

In short: you need a fresh MacPorts (1.6, I had 1.3.something):

$ sudo port selfupdate

, and then, wipe your current libxml2 (along with any dependent libraries, MacPorts will tell you which):

$ sudo port uninstall libxml2@2.6.23_0

(You might not need the version number; I had tried to get the new libxml2 before this so there was an Activating libxml2 2.6.31_0 failed: Image error: Another version of this port thingy going on.) Then just:

$ sudo port install libxml2 libxslt

and you'll be ready to do the final (after removing any misbehaving lxml (compiled with the old libxml2) from site-packages) :

$ sudo easy_install lxml

But why?

Because lxml 2 is a marvellous thing for churning through XML and HTML with Python. There was always XPath and XSLT, C14N in lxml 1.x too (admittedly also in 4Suite as well; Python has had strong XML support for many, many years). But in lxml 2, you also get:

• CSS selectors
• much improved "bad HTML" support (including BeautifulSoup integration)
• a relaxed comparison mechanism for XML and HTML in doctests!

And more, I'm sure.

So, why all this Yak Shaving tonight? Just this afternoon I put together an lxml and httplib2-based "REST test" thing. Doesn't sound too exciting? I think it may be, since I use it with the old but still insanely modern doctest module, namely its support for executing plain text documents as full-fledged tests. This gives the possibility (from a tiny amount of code), to run plain text specifications for ROA-apps with a single command:

Set up namespaces:

 >>> xmlns("http://www.w3.org/1999/xhtml")
 >>> xmlns(a="http://www.w3.org/2005/Atom")

We have a feed:

 >>> http_get("/feed",
 ...     xpath("/a:feed/a:link[@rel='next-archive']"))
 ...
 200 OK
 Content-Type: application/atom+xml;feed
 [XPath 1 matched]

Retrieving a document entry with content-negotiation:

 >>> http_get("/publ/sfs/1999:175",
 ...         headers={"Accepts": "application/pdf"},
 ...         follow=True)
 ...
 303 See Other
 Location: /publ/sfs/1999:175/pdf,sv
 [Following..]
 200 OK
 Content-Type: application/pdf
 Content-Language: sv
 [Body]

That's the early design at least. Note that the above is the test. Doctest checks the actual output and complains if it differs. Declarative and simple. (And done before, of course.)

Ok, I was a bit overambitious with all the "ROA"; this is mostly for checking your content-negotiation, redirects and so on. But I think it'll be useful for a large system I'm putting together, which will (if things work out well) at its core have a well-behaved Atom-based resource service loaded with (linked) RDF payloads. Complete with feed archiving and tombstones, it will in turn be used as fodder for a SPAQRL service. (Seriously; it's almost "SOA Orchestration Done Right", if such a thing can be done.) But this I'll get back too..

(.. I may call the practice "COURT" — Crafting Organization Using Resources over Time. It's about using Atom Entries as manifests of web resources with multiple representations, fully described by RDF and working as an over-time serialized content repository update stream. (Well, that's basically what Atom is/works perfectly for, so I'm by no means inventing grand things here — grand as they IMHO are.)

Cutting Edges

For all of you out there writing droves of RDF, churning out endless amounts of Notation 3, and doing it all in Vim.. (we must be in the millions I'm sure), be sure to check out my little RDF Namespace-complete — giving you RDF Vocabulary/Model Omni-Completion for Vim 7+. Requires Vim compiled with Python and RDFLib.

On a related note, I've been using RDFLib a lot lately (around the clock as it seems). It's an honor to contribute in an ever so small way to it, and it's great to see it progressing so well (it's been mature for years if that needs to be said). The SPARQL-support is working, and along with Chimezie Ogbuji's FuXi there is now an even stronger Python RDF platform to work with.

Within the Outer Space

While closures are very interesting and useful, they may be less of a wise tool when implementing an extendable toolkit. The first version of Oort used them extensively in the oort.rdfview package. My code was quite illegible and obviously a case of failing to be "clever". A more "properly dressed" class-based approach was (as I honestly felt even when I first stepped off the beaten path) evidently much more easy to maintain. Still, state change is less evident in the flat representation of a class, which doesn't give a structural hint of the intended temporal aspect. Using stuff like IllegalStateException doesn't alleviate this much at all (it's a "formally polite design" though). The impression of "loading" a state and bundling a nice nest of code to unfold at each state change is a quite powerful aspect of closures, I guess.

So I thought, let's do that in a somewhat declarative manner.. And thus I (just now) sketched up a way to declare a state change as an hierarchical closure nest which must be unfolded in the proper sequence:

class stateflow(object):
    def __init__(self, start_func):
        self.start_func = start_func
        self.steps = []
        self.reset()
    def start(self, *args, **kw):
        self._started = True
        return self.start_func(*args, **kw)
    def next(self, step_func):
        self.steps.append(step_func)
    def __getattr__(self, name):
        return self.__dict__.get(name) or self._get_next(name)
    def _get_next(self, name):
        if not self._started:
            raise BadStateError("Not started.")
        step_func = self.steps[self._at_step]
        if step_func.__name__ != name:
            raise BadStateError("Must call %s before %s."
                    % (step_func.__name__, name))
        self._at_step += 1
        return step_func
    def reset(self):
        self._started = False
        self._at_step = 0

class BadStateError(LookupError):
    pass

#== Let's test it ==

@stateflow
def query(database, subject):
    @query.next
    def with_query_data(predicate):
        @query.next
        def run_query(context):
            return database[subject][predicate] % context

database = {'s': {'p': "value in %s"}}

query.start(database, 's')
query.with_query_data('p')
assert query.run_query("context") == "value in context"

query.reset()
query.start(database, 's')
try:
    query.run_query("context")
except BadStateError, mess:
    assert str(mess) == "Must call with_query_data before run_query."
else:
    assert False

query.reset()
try:
    query.run_query("context")
except BadStateError, mess:
    assert str(mess) == "Not started."
else:
    assert False

Pretty interesting, I guess. Won't be putting that into my toolkit just yet though (if ever). So, dear closures, it was a nice experience, but I just have to let some of you go. Till next time, keep it simple!

(Oort 0.2 is on the way, by the by. I oort to create an introductory tutorial for it.. And a logo. Yes yes, of course; without a logo, what kind of a toolkit would it be?)

Think Outside The Box

I just realized it. The reason for the increasing popularity of dynamic languages may just be:

• Some years ago, many developers sat in front of humongous 19" CRTs and about 1600x1200 or whatnot.


• Nowadays, most of us sit with sleek notebooks with quite the lesser screen size and resolution.

Thus, it's much more of an annoyance that we have to scroll through rows and rows of boilerplate and wrapped lines of type declaration harnessing. It simply has to go.

(Nevermind productivity or the fact that these new tiny things pack enough CPU power to make DNA sequence processing a BLAST (pardon the pun).)

All About the Package

I'm currently in the process of cleaning up and packaging Oort - a Python-based thingy used for making web apps driven by RDF graphs (there will be more where that comes from, rest assured).

Following is my little list of good starters when doing Python package development.

Most obvious: take a good look at setuptools (especially the concepts and usage of eggs, easy_install and pypi).

For testing, Nose is my one-stop-shop for now. Avoids the articificality of the xUnit-API; uses automatic discovery (with filtering); finds doctests if desired; uses coverage.py if available.

It's probably always wise to follow the practises of the most modern and well-renowned packages. I think Paste is a good example of this.