Hey ./manage.py runserver, Y U No Work?

Recently I was working on a Django-based project that integrates with Android and Embedded Linux devices. One day a coworker called me and asked about an error he had encountered. It seemed like he was not able to execute ./manage.py runserver to any port he had previously used. He was careful to use Control + C to stop the server, but for some reason the port was held onto by something. Looking at the output of “ps aux” was not showing any running processes using ./manage.py.

Who’s using that port?

Next instead of looking at processes, I wanted to see what was holding onto that port.  I ran the following command:

netstat -lpn | grep 8000

The output gave me the answer:

tcp    0    0 127.0.0.1:8000        0.0.0.0:*      LISTEN    15902/adb

So ADB was the culprit. A simple kill -9 to the process ID and the issue was fixed (temporarily). The issue keeps occuring, so maybe something in my code may be keeping ADB holding onto that port. I’m using the subprocess module to interact with ADB. I tried adding a call to terminate from the result of my subprocess.Popen call, but that ended very badly and had no effect on the hanging port. I’m sure when I have time to dig deeper (read “After My Looming Deadline”), I’ll try to write some tests to explore this issue more thoroughly.

I’ve been working on a Django-based project since October 2010 that includes some integration with Scrapy. Scrapy is a web scraping framework for Python and due to its design inspiration, Django developers will be quickly familiar with it. I have installed Scrapy no less than 10 times at this point and decided it was time for a script. This post is more of a reminder to me for the next time I install Scrapy, but if it works for you as well – great!.

This script works for Ubuntu 9.10, 10.04 and 10.10.  If you’re on an older release, you’ll need to install some dependencies first and then follow these instructions.

sudo apt-get install curl
curl -s http://archive.scrapy.org/ubuntu/archive.key | sudo apt-key add -
DISTRO=$(lsb_release -cs)
echo deb http://archive.scrapy.org/ubuntu $DISTRO main | sudo tee -a /etc/apt/sources.list
sudo apt-get update
sudo apt-get install scrapy-0.12

Have you ever edited your models.py file and shortly after when trying to see the change list for a model in the admin site you received this error?

Caught an exception while rendering: coercing to Unicode: need string or buffer, NoneType found

I did today and it was one of those “bonehead” moments. You know, the one where you’ve been coding for a bit and for some reason make a simple error that leaves you thinking “I’ve done that exact same action a hundred times, why would I get an error this time?”, when you realize you made a really simple mistake. The answer was quick and easy. It was not occuring on the new model form, only on the change list so I opened my models file and looked over the model in question and then realized my typo. I had forgot to add “return” to the __unicode__ method on my model. So if you get this error, make sure that you’re returning something and that its really unicode.

I was working on a template last night to handle a generic view. I started by going into the console (managey.py shell) to poke around at the model I wanted to use. From that session, I copied the following snippet to use in the template.

project.projectimage_set.all()[0].image.url

I added this to the template and it looked like this.

{% for project in object_list %}
<div class="project">
   <div class="project-image">
      <img src="{{ project.projectimage_set.all()[0].image.url }}" alt="" />
   </div>
   <div class="project-info">
      {{ project.name }}
      {{ project.link|urlize }}
      {{ project.description }}
   </div>
</div>
{% endfor %}

When running the template I received the error: “Could not parse the remainder: (0)”. I hadn’t seen this before and searched Google to get the answer. It didn’t look good on the first search. The answer was repeated in different results, originating from a Google group. The answer given (and I’m paraphrasing) was “Django templates don’t use Python code. You’re doing it wrong. Go write a template tag or custom method.” I read through the link a few times and couldn’t find the answer I wanted to hear. I figured this was common, and I thought a web framework for people with deadlines would not require me to write too much code to get to a property.

I went back to the documentation and looked around and still didn’t find anything that showed me how to access the properties further down on the related model so I went back to Google. This time I found an answer!

I’m not going to put on like I know what’s going on here. I can think about how somewhere underneath in the template language there’s an implementation of __getattr__, but I didn’t go and research it. It’s almost 0200 and sleep is calling. Because it was difficult for me to find, I thought I would just post this and maybe it would help the next guy stumble upon removing the () and [ ].

Here is the working code:

{% for project in object_list %}
<div class="project">
   <div class="project-image">
      <img src="{{ project.projectimage_set.all.0.image.url }}" alt="" />
   </div>
   <div class="project-info">
      {{ project.name }}
      {{ project.link|urlize }}
      {{ project.description }}
   </div>
</div>
{% endfor %}

This was a great talk where Malcolm Tredinnick spelled out what he’s looking for in Django patches and what to avoid if you don’t want your patch tossed out.  Although his talk was targeting the Django project, he really dispensed good advice for anyone maintaining software regardless of project, language or platform.  The room was completely packed!  All of the chairs and couches we’re filled and people were sitting on the floor and standing at the sides of the room.

You’re doing it wrong if:

• the word “print” is in your patch
• you don’t have a test (fail before, pass afterwards)
• you think “PEP 8″ is an energy drink (PEP 8 is the style guide for python)
∘ make your patch look like django style code

Create patches by running diff from the top of the repository to get all changes.  The svn add command works locally even if you don’t have rights to commit to the repository so files you add will be added to the patch.

“Code style is opinionated and yours doesn’t matter!”

What he’s saying here makes sense in any open source project.  There’s a style defined and you should not deviate from it no matter how much you like your own special coding habits.  You might get away with minute changes, but don’t push it.

“Read the contributing document.”

This seems to be simple and common sense.  Skipping over information someone has left for you is a recipe for confusion.

Some tips given on comments:
• comments should last
• comments should be correct
• comments should explain they “why”, we can already read the how in the code.

Some general tips:
• Fix problems and not symptoms!
• Research is not a four letter word.
• The crowd is smarter than you

And what I believe was his last advice for the talk (at least I stopped taking notes after this) was to contribute and keep contributing.  Even when your patches don’t get in, don’t get discouraged.

It was a good talk and the first time I had heard Malcolm speak.  He was very informative and encouraging.  Overall I liked it and it was one of my favorite talks besides James Bennett and Cal Henderson.

I’m here at the GooglePlex sitting a row away from the stage and enjoying the atmosphere.  It’s a great vibe here.  This does feel different than the Microsoft conferences I have been to.  Some observations so far is that the crowd seems a little younger for the most part and there is a comfortable atmosphere.  Of course there’s free food and drinks, free wifi, a great stage and screen for the presentations and the audio is flawless.  The only complaint I have thus far is the uncomfortable plastic daycare chairs.

Earlier I attended Guido’s App Engine keynote (blog post to follow later), and I’m now attending the High Performance Django talk.  I will try to make a post for each talk I attend.

Check back soon…

Because I can. Because I should. Because cool developers do it. I’m really doing it because I agree with an article from Geoffrey Grossenbach of the Nuby On Rails and PeepCode (actually, the guy does tons of stuff – go check his site out) fame. In it he says,

“I’ll add to this and say that every beginning Rails developer should write their own blog software. It’s a great learning experience and you can try things that aren’t possible with just an app running on localhost. It’s also a great environment for learning without the pressure of a mission-critical app. When you’re working for a client and deploying an important application, you’ll have made all the beginner mistakes on your own time (hopefully).”

It’s somewhat of a personal pride thing, as I want readers that come to my site to know that I took the time to build my own blog engine. It was a great introduction to Django and provided me a means to applying the knowledge I have gained from the tutorial and other reading.

So far I’ve done three 45 minute sessions during lunch and I have the Articles, Categories, Links, and Announcements working.  I’ve been keeping track of the time and once I’m done I’ll provide the stats on time and effort along with the source.