Aqsis Developer Meeting - 20071125 (21:30 GMT)

Paul Gregory has been doing preparatory work for the Beaker project, mainly setting up the test suite. Also discovered and eliminated a couple of bugs while doing so. Chris Foster has been working on the texlib work, focusing on the re-implementation of the IqTextureMap interface. Separation of texture sampling and loading, reducing the load on the interface. Implementing Monte Carlo and EWA filtering.

1. Document the repository/branch/build strategy on the Wiki - renderguy
2. Expand on the documentation tasks in the discussion tab of the main documentation wiki page - renderguy
3. Get tiled TIFF input working (aqsistex) - c42f
4. Complete test content specification for Beaker, and prepare benchmark results from initial branch point. - pgregory
5. Prepare Linux based test machine for Beaker - pgregory & tucks

There has been some progress on the Beaker project setup. We will discuss the current status of the test content suite, the newly created RTS task type SpeedTest, and the next steps of the project.

Paul has added various tests to the test suite, trying where possible to make sure that the tests focus to some degree on quality while remaining true to the importance of testing certain features. The reason for this is that the Beaker project may end be publicised to gain some interest, and if it looks good, all the better. One more test is needed to test motion blur, Paul will address that during this week.

To handle image attributes (arbitrary data which may be stored in an image file header) you need some kind of hetrogenous container, and ideally a typesafe method for extracting data from that container. I'll discuss a little about how my new design solves the associated issues, as I think it's an elegant and somewhat general technique.

Chris has been working on a neat general system for handling 'tag' data in the texturing system. Using boost::any as the basis, but a very clever trick in the interface to allow compile time checking of tag types and values. This ensures type safe operation, and ensures better performance. Read the logs for more information, and hopefully, Chris will be looking to document the process on the wiki at some point for reference.

1. Document the repository/branch/build strategy on the Wiki - renderguy
2. Expand on the documentation tasks in the discussion tab of the main documentation wiki page - renderguy
3. Complete test content specification for Beaker (motion blur), and prepare benchmark results from initial branch point. - pgregory
4. Prepare Linux based test machine for Beaker - pgregory & tucks
5. Document the proposal for bucket overlap subproject in Beaker - pgregory
6. Implement EWA filtering in texlib - c42f
7. Implement support for all sampling options passable via the texture() calls - c42f

The complete transcript of the meeting can be found here.