Software Projects

Open source projects that I have contributed to (patches, docs, bug reports, etc.):

Ohloh profile for dlatornell

My Bitbucket account is https://bitbucket.org/douglatornell/ and my Github account is https://github.com/douglatornell/

Personal Projects That I Have Released

Blogofile and blogofile_blog

I am the lead maintainer for the Blogofile project, a role I assumed when the project's creator lost interest in continued development of Blogofile, but the user community refused to let the project die. Blogofile is a static web site builder with roots in the creation of blog sites. It provides a way of building dynamic feeling web sites (like blogs) that don't require a database or other complex computing infrastructure. Blogofile sites can be served from most hosting providers. Because the sites that Blogofile produces are composed of static web pages they avoid most of the security vulnerabilities of dynamic site tools like Wordpress. Blogofile sites require minimal maintenance, freeing their owners to focus on content creation and other fun things in life.

When I took on the lead maintainer role Blogofile was in the midst of being transformed to a core and plug-ins architecture. I completed enough of that work to get to the 0.8b1 release of Blogfile and the blog creation functionality as a reference plugin implementation, blogofile_blog. Significant features of that work were:

  • Unification of the codebase for Python 2.6, 2.7 and 3.2 (no 2to3 or 3to2 conversion required).
  • Moving the project documentation source files have been moved into the project repository. They are built and rendered at http://docs.blogofile.com/ thanks to the readthedocs.org service.
  • Refactoring of the configuration system, and the site initialization syntax and functionality.
  • Improved Unicode handling in several areas, with the help of Blogofile's increasingly international user community.

SOG Coupled Biology & Physics Model of Deep Estuaries

SOG is a research model developed by Dr. Susan Allen's coastal oceanography group in the Department of Earth, Atmosphere and Ocean Sciences at the University of British Columbia. The model is a vertical, one-dimensional, coupled physics-biology model for deep estuaries. It is a local model of the lower trophic-level ecology and carbon/oxygen/nitrogen cycles that is based on a simplified, but accurate, physical model of coastal regions. SOG is being used to investigate the processes that determine primary productivity and carbon fluxes.

If you are interested in the model check out Susan's publications, and if you are interested in collaborating on the code, contact Susan and maybe something can be worked out.

I did a major refactoring of the codebase during our sabbatical in 2006 and I continue to provide software engineering support for its development, and occasionally work on improvements and new features. I maintain the SOG version control repository, and the SOG buildbot cluster that we use for regression testing.

SOG is also one of the concrete, non-trivial, examples that I use to think about and experiment with automated software testing in the context of scientific computation. Sadly, apart from the buildbot regression testing, I haven't come up with anything that makes me happy enough to share.

Strait of Georgia Bloomcast

In late 2011 we started running SOG as an operational model to predict the first spring bloom in the Strait of Georgia. The model is run daily, driven by a combination of past actual forcing data and averaged future forcing data. For a given day's run the meteorological and river flows forcing data are collected from Environment Canada web services. Those data typically lag the run date by 2 to 4 days. The forcing data for the period from the end of the actual past data to the end of run are taken from 3 time series collections:

  • Averaged values over the 42 year period from 1968 to 2010. Those years were the subject of the spring diatom bloom hindcast study reported in Allen and Wolfe, 2012 [1].
  • Data from 1992/93, chosen because 1993 had the earliest predicted bloom in the Allen and Wolfe, 2012 hindcast study.
  • Data from 1998/99, chosen because 1999 was tied for the latest predicted bloom in the Allen and Wolfe, 2012 hindcast study.
[1]Allen, S. E. and M. A. Wolfe, in press. Hindcast of the Timing of the Spring Phytoplankton Bloom in the Strait of Georgia, 1968-2010. Prog. Oceanogr. Accepted for publication.

The bloomcast results are available at http://www.eos.ubc.ca/~sallen/SoG-bloomcast/results.html That page is updated daily between 1-October and 30-April. Between 1-May ad 30-September it shows the results of the previous spring's bloomcast runs.

In the course of the 2011/2012 bloomcast prediction runs it was found that the correlation algorithm used to transform sky descriptions to cloud fraction forcing values was resulting in lower than actual cloud fraction values. That caused the predicted bloom date to be too early. That issue has been addressed with a new cloud fraction mapping that is based on an analysis of weather descriptions and cloud fraction values reported by Environment Canada over the 2002 through 2011 period.

The bloomcast driver has recently been modified to take advantage of multi-core processors by running the 3 model calculations concurrently.

The bloomcast driver is written in Python. It uses the requests library to collect the forcing data from the Environment Canada web services. The results page is rendered from a Mako template. It incorporates profile and time series graphs that are created using matplotlib and rendered as SVG images. The HTML and CSS are based on HTML5 Boilerplate.

SOG Command Processor

Another SOG-related project that I have recently implemented is a command processor wrapper written in Python. The current release of the tool provides an automated input file editing feature that simplifies the preparation of large collections of runs for sensitivity analyses, genetic algorithm parameters optimization, etc. This has enabled SOG users to craft batch run programs that execute in excess of 100 runs in 24 hours on a cluster of multi-core Linux workstations. Future plans for the SOG command processor include automating the execution of batch runs, initially on a per-workstation basis, but ultimately providing queue manager and worker functionality to automatically distribute batch runs across the workstation cluster.

Proprietary Projects

I have developed a number of software tools for my employer, Nordion:

  • I-123 Datalogs
    A data logging system that monitors the operation of 3 production systems that produce I-123 radio-isotope via proton bombardment of enriched Xe-124 gas. Sensor data and active device states are logged from PLC in a MySQL database. A PHP web application provides trend graphs and status reports to users.
  • IsoInfo
    A web application that provides planning, prediction, and data logging tools for production of Sr-82 via proton spallation of natural Mo targets. An irradiation model and web application implemented in PHP provide tools to plan irradiations to meet production demands, to analyze irradiations in progress, and to provide a basis for production yield calculations. Beam on target data is collected nightly from the TRIUMF 500 MeV cyclotron data logging system, stored in a MySQL database, and used in conjunction with the planned beam schedule to drive the irradiation model.
  • ZSM
    A data logging system that monitors the state of a radio-pharmaceutical manufacturing cleanroom suite. Temperature, humidity, differential pressure, and non-viable particle counter data are logged from a collection of sensors by a LabVIEW application and stored a MySQL database. A TurboGears web application provides trend graphs, alert logging and investigation, and sensor calibration management to users.
  • Minerva
    A quality management system web application that manages quality system form data in conformance with various standards for pharmaceutical manufacturing. Minerva provides form creation, routing, and tracking functions with electronic signatures, as well as quality system performance metrics, and full-text searching. The web application is built on the Pylons framework, and CouchDB is used to provide the data persistence layer.