For a Django project, a metric like the number of apps or models could be a better metric

The number of SLoC is a flawed metric for any piece of software: it abstracts many of the underlying complexity present in the craft of creating software. Depending on the skills and experience of the developer(s) writing the code, the exact same functionality can be written with a very different SLoC number count.

How do we measure complexity ? Is complexity rather than the raw physical size of the codebase a better metric for the “size” of a piece of software ?

Of course there are algorithms to measure such things as cyclomatic complexity, but that does not evaluate the inherent complexity of the software, but the complexity within the logical paths coded in the application. More often than not, high numbers for such a metric is just a sign that something is wrong with the code : clean code will almost always tend to have a lower score for these analysis.

The repository is at http://hg.peach3.nl/

I’m planning on making a real release for django-pdfviewer someday. Currently its still a bit rough in some places.

You mention django-pdfviewer in this post. I cannot find the sourcecode in the trac for peach³. Maybe you can point me to where I can find it?

Would be really nice! Thanks.

It’s a system for managing real estate information, coupled with customer sites in a SaaS fashion (each realtor gets their customizable site inside a subdomain). There’s also a business network where realtors can partner with each other and share information. Something a bit different, as the target market is Brazil, and the real estate market is diverse here.

We are making using of a bunch of different Python and Django related tools. Full text search ORM + Tsearch2, GeoDjango + PostGIS, South, Pinax modules, ImageKit, Python image processing libraries, custom middlewares for SaaS, (many) caching frameworks for memcached and others, Amazon AWS API libraries, django_extensions jobs for billing and cleanup, Fabric deployment scripts, etc. All being developed and improved in record time.

It’s really awesome how easy is to integrate stuff with Django, the framework barely stays in your way (as opposed to my past experiences with other frameworks, namely Rails ). For example, integrating GeoDjango with TSearch2, and being able to search properties using capabilities from both (coordinates and full text indexing) in the same QuerySet was a breeze. Definetly a success case for Django.

Let’s see what the statistics are according to them.

Satchmo has 144,000+ lines of code+ markup or 71,000+ for code only.

I can imagine longer running projects may have a substantial amount of javascript, flash or other non-python code. These should be counted as well since they are part of the working project.

What about documentation? In my view, lots of code with virtually no documentation is rated lower than a small, clear and well documented project.