Made the code easier to follow :
1. suppressed the mjd -> year transformation to access epoch
T is now computed directly from mjd - J2000.
2. conversions limited to arcsec -> radians needed by the precession formulae
3. increased PI resolution, added TWOPI, radarcsec and arsecrad constants in astro.h
added a generic computation function that does
both backward to 2000.0 and forward from 2000.0
depending on one parameter dir which indicates direction :
if dir <0 computes transformation from equinox to 2000.0
if dir >0 computes transformation from 2000.0 to equinox
This avoids code duplication, noticeably for the
transformation formulae.
read ira as RA (UCAC RA at epoch epu)
and dcg as DEC (UCAC DEC at epoch epu)
instead of the GAIA values rag and dcg.
(http://tdc-www.harvard.edu/software/catalogs/ucac5_format.html section c)
The reason for the switch is that rag and dcg are
given at epoch 2015.0 ; and it proves hard
(at least I failed) to figure out exactly what is taken into account here ;
the xephem internals apply precession and other corrections
later, so there might be some ambiguities here. And indeed,
there are significant differences between expected results
when compared to those obtained in the vizier database interface.
The issue does not even pop with ira and dcg, based on ucac epoch,
very similarly to what is done in ucac4.
Now xephem shows results (tested on a few stars) within 0.01 arcsecond
of the data from the vizier interface.
1. there was a 4 byte shift error in decoding raw records for rawpmra
and rawpmdec.
2. Proper motion management clarified. readu5hp() is unnecessary as far
as I can tell, proper motions are included in raw records.
In crack5(), raop->f_RA and op->f_dec are referenced to J2000, and
are later processed to account for proper motion up to the current
epoch (circum.c:obj_fixed())
Successfull proper motion check against high proper motion star
HIP74995 (>1arcsec/yr)
3. some code cosmetics
Previously, a wide Sky View window would be resized to a square after a
DSS download completed. From groups.io mailing list patch
“xephem-4.1.0_skyviewmenu_sv_getcenter.patch”.
Also, initialize openssl as early as possible. (Brandon notes: this
breaks compatibility with older systems, but he will follow up with a
patch fixing the problem.)