magpiebrain

Sam Newman's site, a Consultant at ThoughtWorks

Posts by samnewman

Since lambasting someone for a less than fair Django and Rails comparison, I’d planned to do my own. To start with, my comparison is based on several days worth of Rails development and about a weekend’s worth of playing with Django, using the (currently four part) tutorial.

I’ve had this reviewed by both Rails and Django developers, but if you think I’ve misrepresented anything or left anything out then feel free to leave a comment and I’ll make sure to put it right.

Background

Rails is framework which emerged from the development of a web application, Basecamp. Django was emerged from the development of the highly successful Lawrence Journal World. Both applications were written in situations and organisations that historically would have ended up being written in Perl or PHP – but ended up being written in Ruby and Python respectively.

The fact that both frameworks were the result of development of specific applications means that both frameworks are focused on supporting those specific types of application. As a framework Rails has been around for little over a year, whereas as a framework in its own right django has only been public for around two months. Neither has yet made it to version 1.0.

Languages

On the face of it, the biggest difference between the two has been the choice of development language. Outside of Japan Ruby has made little impact despite having been around as long as Python – a colleague of mine once said that the number of Python programmers in London probably outnumbered the number of Ruby programmers in Europe and the US and he’s probably not wrong. Anyone having to consider which of the two frameworks is right for them is inevitably going to have to think about the task of hiring people and this may have an impact on their choice.

Ruby does seem to have some more powerful language features – it has slightly better closure support than Python (although perhaps not enough to be a differentiating factor) and the extremely useful mixins. Ruby’s OO support also feels much cleaner. Both support inheritance (Python supports multiple inheritance, which Ruby’s mixins more than make up for). But the requirement that self be passed in to all python instance methods drives me potty.

On the other hand Python has a much cleaner syntax (which given its background is no surprise) – more than once I’ve found myself using a third type of parenthesis in the same Ruby method quite validly, only for me to look back at what I’ve done and feel that I’d regressed to the bad old days of Perl. It would be nice to think that one day a language will come along which manages to add powerful features without making the code look like it’s been rendered with the wrong character encoding.

Ultimately comparing two languages that are so close (they’re certainly more similar to each other than Java or .NET) is going to come down to personal choice or a specific requirement. But for more indepth comparisons you’ll no doubt find many out there on the web.

Requirements and installation

Installing both Rails and Django is fairly easy once you’ve got the relevant interpreter installed. Ruby can make use of the excellent Gems system, which is something Python (and to be fair most other languages including Java) would do well to mimic.

Both also come with their own webserver for development purposes, which is great from a developer’s point of view. There should be anything stopping you from getting both up and running fairly quickly

Deployment options

Both Python and Ruby suffer from a similar global interpreter lock problem – basically neither the Ruby nor Python interpreters are completely thread safe. When running multi-threaded applications a lock has to be acquired by a thread before it can access any objects. That means that both Ruby and Python scale by using multiple instances of the interpreters rather than the Java approach of using multiple threads – as a result the deployment options are fairly similar.

Django’s current recommended deployment is using Apache and mod_python, although there is no reason why fastcgi couldn’t be used, or even an alternate server such as lighttpd (which is TextDrive’s current approach to hosting Rails).

One important point is that Django currently only supports Sqlite, MySql and Postgres (although support for Oracle has already been submitted to the project), whereas Rails supports virtually all the major relational databases.

Caching

Django’s heritage as a framework supporting large, high-volume websites shows through clearly in the built-in support for a variety of caching options. Rails’ own caching doesn’t seem as flexible. With Django multiple caching backends are supported (in memory, memcached, or DB) coupled with supporting caching at multiple levels – either for a whole site, a specific page or fragment (like Rails), or even a specific object.

Setting up a project

Both frameworks generate your project structure for you. Django actually has separate concepts of Project and Application (of which more later). Rails goes a little further – it also generates a build file and stub tests (I won’t call them Unit tests as Rails does, because they’re not). There is nothing to stop you testing Django of course – it’s just as testable as Rails – however it would be nice to get some guidance on how to do it. My feeling is many people new to testing are more likely to do it with Rails because it’s built right in and it’s well documented.

Adding models

In Rails, you define a table that represents your model, and then generate the stub code to create a Rails model that binds to the table. You are then free to build upon the stubs. With Django, you define your model in Python code, and have Django generate (and even execute) the schema for your model.

The fact that your Python code can be the canonical reference for your models means understanding can be simplified. It can seem odd to be adding code to a Rails model and not know exactly what attributes your model has without referring to the database schema. The fact that Django can generate the SQL from you model also leads to flexibility in development – for example a developer can use Sqlite which is fast use and trivial to setup, but deploying onto MySql or Postgres is easy as you can regenerate the SQL for your target database from the Python code – that the SQL for both databases might need to be different is hidden from you.

Early on your model will be constantly changing – I’ll often find myself writing some Rails code, need to add a new attribute to a model, have to change my (DB-specific) schema. With Django I change my model and regenerate the SQL. The only minor annoyance is that there is no equivalent of a model-specific SQL refresh – something that could drop a specific model’s tables and regenerate the needed SQL automatically. Instead you have to have the admin script dump out the SQL to the command prompt and run it in yourself.

Django’s generation of schema also extends to creating lookup tables for many-to-many relationships.

The downside to having Django generate the SQL schema is that there is more work involved in supporting databases – currently only MySQL, Postgres and Sqlite are supported, although I’d be highly surprised if support for other databases wasn’t on the way.

Django does seem to offer the ability to reverse-engineer models from an existing schema, although I don’t have information as to how sophisticated this is (for example to what extent it can determine inter-model relationships).

Another thing to note is that unlike Rails Django doesn’t allow you to manage your own transactions (although it is going to be added). I’ve not found myself needing it in my Rails or Django work to date, although as a J2EE programmer not having the ability to control this if I need to is a minor concern for me.

Interaction with the Database

Both Django and Rails have taken the decision to bind models to the database. Methods to save, delete or lookup objects are called on the models themselves. Neither specifically stops you from providing proper separation, but if you do you’re on your own. Going forward it will be interesting to see if either Rails or Django start providing the ability to have a clean repository layer as nicely as they do the current DB access, although it will probably require both of these frameworks to be used in more complex situations than they are currently.

One nice thing Django offers is the ability to write your own SQL in the model to access attributes. Rails does nothing to stop you calling SQL directly, but the fact that with Django you are doing this custom lookup in the same place as the definition of the other attributes (whereas with Rails it would be split between the schema and Rails model) results in something much cleaner.

Projects and Applications

A nice touch with Django, and another example of the environment in which it was developed, is that there is a project in which you can have pluggable applications. This should hopefully lead to smaller (Rails-style) apps being created, reused and integrated easily into a larger Django projects. This was something that was demonstrated wit LJWorld, with new parts being added incrementally without a need to deploy the whole site.

Controllers and URLs

With both Rails and Django URLs are mapped directly to code. Rails automatically generates stub controllers to map to the scaffolding (of which more later), whereas with Django you have to create them yourself.

With Django you use regular expressions to define URL’s that bind to your python code. That makes it much easier to create RESTful URLs – in fact all of the tutorials reinforce this. The fact that you have RESTful URLs happens to make controlling code simpler is just an added bonus.

By default Rails gives you URL’s in the form controller/method – but these can be edited using the routes.rb file to give you proper RESTful (or not) URL’s. Unlike Django’s pseudo mod_rewrite approach, Rails uses Ruby code to handle routing, which depending on your point of view could be considered more readable.

The URL to code mapping can be specified either for a whole project or (more likely) for a single application.

Templating options

Both Django and Rails template using embedded Python or Rails code (like using Java scriptlets in JSP pages). Much more work has been put into Rails’ helper code – Ruby code you can use to generate HTML constructs – its excellent Ajax support being one example. Django does make it fairly easy to swap in other templating options if required (such as Zope’s Template Attribute Language, or perhaps Clearsilver). I’m less savvy on alternate templating options available for Rails – although obviously if you used something else you’d have to do without many of the helper methods available.

Administration and User management

Django’s administration interface is simply excellent. Unlike Rails’ scaffolding views, it is not designed to be viewed by everyone (being an Admin console), but unlike the scaffolding it is perfectly functional and you could imagine making use of it without alteration in a production application.

The administration console is password protected – you can generate a superuser account using the command line admin script. Once you’ve logged in you can add more users and even groups – Django automatically generates permissions for both the admin functions and your models which can be assigned to a group.

You can also add new model instances in the same way as Rails scafolding, but the view is much better. You have nice popups and shortcuts for date and time fields, dropdown lists for one-to-many relationships, and multi-select lists for many-to-many relationships all provided without any developer interaction.

Rails does have a measure of user administration in the form of the add-on salted login hash generator, but despite some hard work by the developers it does seem problematic and nowhere near as slick as Django.

So which is right for you?

I said at the start that the two frameworks were developed as a result of two very different applications – and most of the differences between the two are a result of this. If you are developing a simple (in a domain model sense) application where you want to use Ajax to deliver a great UI, Rails is probably for you. If however you want to develop an entire site with different types of applications within it – then Django’s plugable applications and general approach might be what you’re after. Equally, the better user and administration side of Django favours portal style applications – this is something you’ll have to do yourself if you want to use Rails.

There is scope for both frameworks to learn from each other, and I still have a question around how well Rails will scale (in terms of the size of the site, not number of hits) – Django has already proved it is capable of this with LjWorld.com. Likewise it will be interesting to see if Django can or even wants to make the use of Ajax as simple as it is with Rails (although work is already progressing along these lines), or if the more powerful language features of Ruby will prove the deciding factor.

Personally, given my own current projects I’ll be spending some more time with Django – but all that means is that it is one more tool that will sit alongside Ruby, Rails and J2EE in my particular toolbox. And just like Java web frameworks before, I fully expect to see more and more web frameworks based on dynamic languages (such as Nitro or SeaSide) start to gain more visibility and yet more options. It seems we’re going to be spoilt for choice.

It’s been a while since I’ve had to unleash the angry kitten, but it seems that the OpenAdaptor developers have rather inadvisably irked me, to the extent that I’ve withheld the kitten’s food for two days now, to make the ensuing carnage all the more…well, carnagy.

To start with, I like the concept and some of the architecure behind Open Adaptor. It’s a fine arhictecture – it’s no suprise that plenty of other APIs out there have come up with similar solutions to the same problems. The JBI spec obviously owes a lot to OpenAdaptor. That said its implementation frequently makes me want to have a few minutes with the developers, a pair of pliers and a blowtorch.

Creating sinks, pipes and sources

Let’s start with the fact that to create anything you’re pratically forced to use property files. No Inversion Of Control for you my son, oh no – here we have a special init method, which is not a constructor because…well, it’s called init, its enforced by the API, and it takes a Properties object. The only benifit over being able to construct objects the way you want seems to be it made it easy for the developers to use property files (I hope for the next version they at least look at spring). That makes tesing so much fun. Please note the sarcasm.

JARs not built using debug information

Yes, you heard me right. Here was I, thinking we lived in the age of high-speed internet connections. But in fact it seems internet bandwidth is so tight that the developers considered providing debug information so we could work out for which odd reason an Adaptor fails a luxury.

Errors that are not errors – or might be

So when your adaptor fails with no clear reason why – well, what should you do? That’s right, turn on debug – chances are a debug message will detail why your adaptor failed. Even better, sometimes the error message will be vague, for example the one that tells me that either my JNDI server may be unavailable, or that it is there but the thing I want isn’t, or that it is there and the thing I want is there, only I might be missing a class, only it won’t tell me what class that might be. That is an error message that can be really fun to work with. And I don’t mean the good type of fun either, I mean the achingly irritating kind of fun that can only result in me hurting my foot kicking office furniture.

A nice idea crippled by an allergy to interfaces

OpenAdaptor has a rather sensible and useful (sounding) feature called message hospitals. Normally, if a message causes an exception the app shuts down. You can however choose to configure a message hospital which will capture all messages which throw an exception – but only an exception of a given type. However by “type” OpenAdaptor means “class” – you have to throw a checked PipelineException. This is bad enough – I now have to have a top-level catch throwable clause that rethrows PipelineException – but the constructor doesn’t even take a root cause. Quite why a simple ExceptionType interface couldn’t be used I have no idea.

Conclusion

OpenAdaptor has a sound premise, but it’s implementation needs bringing up to date. A full rewrite in the form of OpenAdaptor 3 is underway – get involved.

Please note, there are:

  • Lies
  • Damn Lies
  • Statistics
  • Bloggers
  • Fox News

    In decreasing levels of accuracy and balance. Of course this site is the exception that proves the rule…

    That means the next time you hear some blogger raving about something, for the love of $Deity consider that you might not be suckling at the teat of truth, and perhaps instead the putrid orifice of lies.

I’ll be at Open Tech 2005 on Saturday, as it looks interesting and its damn cheap (five of your UK pounds). Topics include blogs, copyright concerns, webservices, and the offical launch of backstage.bbc.co.uk’s developer network – aimed at providing ways to open up BBC’s content so Joe public can program a webservice for it or something. Most importantly it starts at the highly civil hour of 11:30am.

If you’re coming along I’ll be the skinny geeky guy with an under-developed social life and too many gadgets. I should be easy to spot.

Way before I started looking at a cruise control plugin for trac, Tammo van Lessen had been developing one of his own. Tammo’s solution uses XSLT to parse the cruise control logs – and it looks very similar to the cruise reporting application as a result. The big added benifit of course is that it integrates with trac (so no need for a servlet engine to display your logs). The current plan is that myself and Tammo will try and merge our efforts at some point, but until then I recommend you have a look at Tammo’s excellent work if you’re currently using trac and CruiseControl.

It’s OK everybody, there is no point comparing Django and Rails, Obie has it sorted:

There’s one problem dude… No matter how good Django is, it isn’t written in Ruby.

You see, that’s what so great about Obie – he can deliver a balanced, reasoned comparison without letting language politics get in the way.

Package dependencies in Pasta

Package dependencies in Pasta

Today on my current project I had to spend some time creating a client JAR file for use by downstream systems. I needed to keep the JAR as small as possible, with (more importantly) no external JAR dependencies.

There are tools out there for dependency analysis, such as IBM’s Structural Analysis for Java (aka Smallworlds), Compuware’s Pasta or the open source JDepend. As nice as these tools are, they are for analysis and the gathering of metrics. As I’ve mentioned before, unless ‘good metrics’ are in some way enforced by the build process, it becomes very easy for even automatically gathered metrics to be ignored.

JDepend can be run via Ant, but just like Emma or Findbugs the information gathered is used for reporting purposes – it is not capable of failing a build because someone has introduced an invalid dependency between packages.

Japan is a tool which comes with an Ant plugin that will fail a build if your code violates the allowed package dependencies. For example, our downstream systems only need to use classes in the client package – I don’t want to have to include any other code in the client JAR file. So in a Japan config file I place the following code:

Missing

Now if any of the classes in client include any other packages in my codebase, the build will fail. It’s important to note that Japan requires that you define all your dependencies at the package depth you define (the package-depth="4<a href="http://www.c2.com/cgi/wiki?AcyclicDependenciesPrinciple" <2>> means that all source code in com.company.app.xxx and below will be checked) – so for example if our gui package depended on util and client, I’d have to add the line:

Missing

By defining these configurations you can quickly discover circular dependencies) – for example if to get the build passing you find yourself defining something like this:

Missing

You know something is up. The other thing I like about Japan is the fact that because I can now enforce sensible package dependencies, I feel better about spending some time cleaning our packages up, safe in the knowledge that we won’t backslide (assuming no-one goes and sticks everything in one giant package of course). There was one little problem I had though – I had to disable transitive dependency checking as it caused a stack overflow error, but I think once we remove our existing circular dependencies that should sort itself out. I still of course like to have tools like Pasta to help me define acceptable inter-package dependencies, but I feel much happier having Japan in the build just in case I start getting sloppy :-).

chirac.jpg

“It’s all gone quiet over there”
“You’re going home in a rather disappointed ambulance”
etc.

This has been a public service announcement from the Smug Londoner Party.

Don’t they know this is the UK? If you’re over here letting off fireworks, you bloody well better be burning the effigy of a Catholic, and it better be November.

Bastards!