Monday 15 May PM VOTable session - 16h00-17h30 FO = François Ochsenbein, MT = Mark Taylor, TMG = Tom Mc Glynn, DT= Doug Tody, AR = Arnold Rots, SD = Sébastien Derrière, PF = Pierre Fernique, TD = Tom Donaldson * Introduction (FO) * News from Applications (VOPlot v1.2.1, VOTable streaming writer, VOMegaPlot 1.0, TOPCAT v2.1) * I/O problems: conservation of GROUP structures (in TOPCAT) * MT explained that GROUPs are lost because current version of TOPCAT forgets the table is a VOTable (it could have been loaded from a FITS file for instance). Future release of TOPCAT will allow to keep GROUPs * VOTable schema * Embedded spaces * MT thinks '&32;' will act as a space and won't glue together a string as "100&32;200". He suggests using '&160;' instead (to be checked) * Recursivity allowed in RESOURCE, GROUP and OPTION. AXIS doesn't seem to support this recursivity * FO not in favour of another schema * TMG : problem has been signaled to the AXIS team. It seems that the recursive character of the OPTION element is disturbing AXIS. Asks whether some applications use recursive OPTIONs. At the time, he cheats with the schema, to make it work with AXIS * MT and all : agreement to do nothing about this point. Best option is to let the AXIS team fix their library. Otherwise, we'll have to change the schema everytime a tool does not act properly * VOTable expressions * Sexagesimal: debate on how to represent sexagesimal values (which unit? "h:m:s", which datatype? double, string). More generally, how to express the fact that numeric values are described as strings * TMG : sexa is evil ! * MT : lot of tables have numbers in sexagesimal * Date/Time : we should recommend to use numerical values as far as possible. * AR : add a specific datatype for time ? --> would be different from the general VOTable philosophy * FO : we must think of FITS compatibility * How to identify that a field contains a date in ISO9601 format? * MT : need a way for the software to know this field is in ISO9601 * SD suggests the use of utype * DT : utype not generic enough * TMG : what about reverse engineering on the field values ? (if this value has this format, then it must be a time in ISO9601) * MT --> reverse engineering possible, but ugly * Agree on unit="ISO9601" format="string" * Special units like "Crab"? Not described in http://vizier.u-strasbg.fr/doc/catstd-3.2.htx * MT : we can use special quoted units, but it must be properly documented ! * TMG : there is a paper about unit standardization (FITS paper?) * utype: proposal of a "a la XPath" syntax refused at El Escorial (Oct 2005) Need to find agreement on a way to write utypes. * there seemed to be an agreement(DT, AR) about the syntax that FO is suggesting. AR will suggest a modification to the example FO has shown * LINK used e.g. in Aladin (A. Rots remarks it could be expressed as XLink) * Very Large Arrays * New CSV serialization (data structure widely interpretable). * TMG : Problem: requires all VOTable users to be able to interpret a new format that brings nothing new * PF : easier for users to turn their data into VOTable * TMG : CSV users do NOT want metadata * most of the audience was against this proposal * Conclusions * agreement on utype syntax * another schema to be produced (change in the form of the schema) * MT suggests to version it 1.11 in order to indicate the purpose of the schema has not been modified * PF : with the new utype syntax (for STC), we should certainly deprecate the COOSYS section. It can't be removed because of backward compatibility