zfitsInfo.txt

[1] 1201.1340v1 - 'A Tiled-Table Convention for Compressing FITS Binary Tables'
[2] Binary table extension to FITS
[3] http://archive.stsci.edu/fits/users_guide - FITS User Guide
[4] http://fits.gsfc.nasa.gov/registry/tilecompression/tilecompression2.3.pdf

* MAYBE: Variable length array are discribed as '1QB' but shoud be described as '1PB' (default proposition by ([1] 2.B) and ([2] 5.11 and 6.A) . '1QB' should be ok by means of [4] 3)
* JAVA: ZTILELEN is not read. TODO do it and check against it maybe. Now it only uses the number of rows in the tile headers as an information about the amount of rows in the tiles.
* JAVA: TODO maybe implement ZCTYPn with other algorithms. Now only supports 'FACT'.
* GENERAL: add ZCTYPn FACT to standard [1] or [4]
* ALL: ZUNITn should contain the SI-unit () and not the datatype.
* INFO: COMPRESSED TABLES in the 'FACT' compression sheme are stored in little endian, uncompressed std bintables in big endian, as described in ([2] 5.x)
* ALL: TTYPEn should be '2K' to describe the catalog in the right format.
* ALL: The readers ignore the TILE id header (necessary ID by [1]), this is because some of the files are writen wrong by the writer.
* ALL: PCOUNT is set to the size of the spezial area plus the size of the heap, but it should only be the size of the spezial area [2]. This is done this way because of an old error in fverify (already fixed, but all files look like this now). Should be fixed sometimes.