As I have explained in previos general comments in dal group, I think the more attention should be given to the rectified = continuum normalized spectra as they are one of the most usefull results for advanced scientific analysis and so are "fully science ready" despite the fact the flux calibration axis does not have units.
Currently I would tell the normalized spectrum is a kind of "filtered" data (2.4.2) as it is a flux transformation" - perhaps it should be mentioned as an example after the words "flux transformation" (e.g. continuum normalized spectra)
In 4.1.1.5 the format application/fits is a way how to support the single line spectrum FITS with CRVAL CDELT values as I have explained in general comments.
It is not wise to use FORMAT=NATIVE for 1D FITS because it is de facto standard for the majority of optical astronomers, and as I understand, if NATIVE data is returned, the client just passes it further without the interpretation. So I suggest to state the FORMAT=application/fits;convention=IRAF-onedspec for such data and to mention explicitely that FITS mean either binary table fits or simple 1D image with CRVAL, CDELT etc.....
In any case it is not clear for the reader that the FITS is restricted to binary tables only (currently) -everybody just implies this.
4.1.2
there is a typo at keyword TIMERES !
sample value of one year in seconds should be something like 31557600
(if mean solar year is meant) and not 86760
(if the 1 solar day was expected it should be 86400 sec !)
I do not undestand 4.1.2.1 Aperture in addition to 2,4,2 Spectral Extraction
"the aperture is specified as a diameter in decimal degree"
Is the raw 2D longslit spectrum a example of datacube ? And can be the specralExtraction service the 1D spectrum extraction of particular object or place ? Typically a rectangular aperture is used for this (1pix wide and height depends on my wishes - e.g. central part of nebula.. so Why only diameter (it implies circular) - what about rectangular like 5E-4x1E-5 as a string - just talking about spectral extraction service ...
In 4.1.2.10 FLUXCALIB is a place to define the normalized spectrum:
is it flux calibrated ? I dont think so. But if it is accepted as relative
flux calibration ? But I may be interested only to find continuum
normalized spectra - how to query this?
I suggest to use the string and insted of yes/no accept "relative",
"absolute", "normalized" or something similar.
In section 7.2 again the application/fits says nothing about the restriction to binary tables only. Moreover in proposed FITS standard http://fits.gsfc.nasa.gov/draft/fits_standard_30.pdf in section G 1.1 below is written that
"As examples, files with media type `application/fits' might contain any of the following contents:" .....
"A primary HDU containing a single image along with keyword/value pairs of metadata."
In the Appendix A I would appreciate another simple example how to return the list of simple 1D FITS from an small observatory archive - It might be the first page people from many observatories would read in attempt to make their simple spectra VO-compatible ....
Best regards,
Petr Skoda