Doug Welch has mentioned the utility of a VOTable observation source plug-in, for example in relation to MACHO data.
Here's the relevant of an email excerpt from Doug on Oct 23: "The columns are...described in the XML file by a set of descriptors which hark back to the original names that MACHO Science Team members (and others who worked with the data) wouid recognize. This is especially true for the "MACHO SoDoPhot photometry descriptors". My feeling is that any code should use the descriptors as column names rather than the word "descriptions" in the first message I sent about this.
Even a routine which only went in and sought out the <DATA><TABLEDATA>
portion of the XML file could easily and correctly parse the lightcurves since they are all plain ASCII.
The MACHO BMAG and RMAG magnitudes are on an instrumental system (corrected to the same zeropoint, of course) but are all negative. This is as it should be. The correction of the photometry to standard bandpasses is not something we should involve ourselves with in VStar. I can tell you all about it, but it has essentially no relevance to time-series analysis.
For longevity, MACHO is one of the best (on-going surveys will pass it, of course) - it provides time series over 8 years for LMC, SMC, and Milky Way Bulge fields."
He also said in an earlier email: "I have been trying to decode the various classes that TOPCAT uses. It looks like VOElementFactory might be an important class: