Thursday, 28 February 2013

Should OGC re-prioritise Quality over Quantity?

There is a constant tension in development between doing less, but doing it well, verses doing more with less attention to detail.
Recently members of the Open Source community have raised concerns about the quality of OGC standards, which leads to reduced uptake and effectiveness of these standards. This should be of concern to sponsors of OGC initiatives, and I'd suggest that the OGC, and OGC initiative sponsors should assess funding priorities, and place a greater emphasis on quality over quantity.
Here is a sample of a few recent comments:
  • Daniel Morissette, board member of OSGeo, blogged: Don't upgrade to WMS 1.3.0 unless you have to, stick to 1.1.1.
  • Chris Holmes, a prominent OSGeo thought leader, recently said: "we've found that in recent years there are even more 'ideas' added to the specifications that have no true production ready working code against them. Past the surface accessibility this is the thing that has become clear to us as implementors - there is less quality control at the core of the standards process."
  • Chris quotes Justin Deoliveria, an experienced implementer of OGC standards as saying "... the geopackage spec makes me want to run for the hills".
In the last OGC OWS-9 testbed, LISAsoft was engaged as part of the CITE compliance program to test the WMS 1.3 client reference implementation. We are strong supporters of testing, but the scope of LISAsoft's recommended testing was substantially trimmed due to lack of budget. Consequently, the level of testing sponsored fell well short of what should be covered by a standards organisation. By acknowledging that testbeds like OWS-9 are vehicles for rapid prototyping, full compliance testing should probably be advanced into another forum.
Here is an exert of what we summarised in our wrap up Engineering Report:
The WMS 1.3 Client testing provided in this testbed is an excellent step forward, and is useful for checking integration between a specific WMS client and specific WMS server, however, the WMS 1.3 client testing falls short of providing comprehensive tests confirming that a client conforms with all WMS servers, under all conditions. As such, LISAsoft considers it inappropriate for the OGC to consider a WMS client to be certified as compliant based upon the level of testing sponsored to date.
For instance:

  • Boundary testing was not sponsored: Can the client send a query which crosses maximum latitude or longitude lines? ... 
  • Exception testing was not sponsored: Would the client perform suitably if the server provided a valid exception as a response? ...
  • Sponsorship only covered testing of integration with one WMS service (which only supported a subset of the WMS spec). Testing the rest of the client's WMS support wasn't sponsored. ...
The recommendations from our report explain steps to bring OGC compliance testing in line with  best practices. However, they could be summarised by recognising that:
  1. Quality of OGC standards is strategically important in reaching OGC goals.
  2. Greater focus on testing is required if OGC wish to reach this level of quality.
  3. Sponsors of OGC initiatives should adjust funding priorities to put a greater emphasis on quality over quantity.


No comments: