BUGS (Oct97)                      xray                      BUGS (Oct97)



NAME
    bugs  --  known  bugs  with  symptoms  and  work-arounds  in current
    IRAF/PROS
    
    
DESCRIPTION
    
    1)   IRAF/QPOE
    Class:  ANNOYING and FATAL - 
    Workaround: YES and mostly automated by use of ROSAT FITS macros
    Symptoms:       obvious
    
    There is an IRAF bug affecting some uses of regions applied to 
    non-sorted (e.g. time-sorted) QPOE files.  If encountered, (the task
    will die with a FATAL ERROR), this problem can be worked around
    by using QPCOPY first and then TIMSORT or QPSORT with region=NONE.
    (NOTE: this is VERY likely to occur when using GERMAN archive data
    which is NOT spatially sorted.  RARC2PROS will correctly sort the
    data, but a manual conversion will not.)
    

    2)   PROS/REGIONS
    Class:  MODERATE and SUBTLE
    Workaround:     YES
    Symptoms:       very subtle
    
    The pixel at the point of a pie slice is sometimes included in the
    region, sometimes not.  It seems to depend on some combination of
    the angle and the x, y position.  The philosophy of inclusion/
    exclusion of pixels in/from geometric shapes, as outlined in "help
    coords", intends that no pixel on a common boundary of two shapes be
    included in both, or excluded from both.  According to that
    philosophy, the central point should be included in exactly one of
    any set of pie slices with that common center -- a very difficult
    requirement to implement in a sensible algorithm.  "help coords"
    leaves the user guessing about what will happen at the vertex of a
    pie slice (and at some other points); the rules described there
    don't cover all cases.  And the fault is not in the rules' being an
    inadequate expression of a clear intention, but rather in the
    intention's being impossible to satisfy with a simple sensible set
    of rules that cover the whole set of shapes we use.
            This can be checked using pllist and sometimes fixed
    by including an additional specifier
            POINT xxx  yyy




                                                                  -page2-

 
    3)   PROS/REGIONS
    Class:  ANNOYING and FATAL - easily worked around
    Workaround:     YES
    Symtoms:        subtle
    
    There is a memory problem in REGIONS that prevents the use of too many
    "accelerated" regions.
            e.g.  ANN 512 512 0 100 n=50
    may fail with a memory error but it can be run in segments successfully
    as follows:
            e.g.  ANN 512 512 0 20 n = 20
                  ANN 512 512 20 40 n= 20
                  ANN 512 512 40 50 n =10
    

    4)   PROS/DATA
    Class:  MILD and MOSTLY ELIMINATED
    Workaround:     YES
    Symptoms:       subtle
    
    There is also a problem with the reference pixel value in the ROSAT
    and EINSTEIN Rev 0 FITS IMAGE files that results in the WORLD Coordinates
    being in error up to 1 pixel.  This error can be avoided (corrected)
    by reading files into IRAF/PROS using the tasks "RFITS2PROS" or
    "RARC2PROS" with the option "correct+".
    

    5)   PROS/QP2FITS
    Class:  MILD restriction
    Workaround:     NO - not yet
    Symptoms:       subtle
    
    The QP2FITS task does NOT work for "rotated" files, i.e. qpoe
    files whose "CROTA2" != 0.  By default all ROSAT and Einstein QPOE
    files are NOT rotated.  However, the output from QPROTATE will be
    a rotated file, not supported by QP2FITS.
    

    6)   PROS/ROSAT/UT time calibration
    Class:  MILD restriction
    Workaround:     NO - not yet
    Symptoms:       subtle -  USER MUST CHECK PERSONALLY




                                                                  -page3-


     The ROSAT spacecraft clock to UTC calibration files are issued at 
     infrequent intervals by MPE, and it's possible that users will 
     encounter data whose times are not included in the current calibration 
     file. Conversion of such spacecraft times to UTC are extrapolations 
     and may not be very accurate. Users should verify that the times of 
     their data are included in the current calibration file by running 
     the task SCC_TO_UTC, which automatically displays the end time of 
     the data file. Updates to the PROS calibration file will be issued as 
     soon as new calibration files are released by MPE. 
     [xray.xtiming.timcor.scc_to_utc]

    
    7)   PROS/DATA
    Class:  MILD documentation error
    Workaround:     YES
    Symptoms:       
    
    The Rev 0 ROSAT/PSPC *_src.tab files incorrectly list the "SRC" entry
    as containing SOURCE+BKGD counts: "SRC" contains the background
    subtracted source counts.
    

    8)   PROS/XSPECTRAL
    Class:   MILD restriction
    Workaround:     NO
    Symptoms:       warning message
    
    A black body model with temperature .01 or less will not work.
    Temperatures this low are applicable to white dwarf stars and will be
    required by some users.  A warning message has been included to tell the 
    user they have reached the limit.  
    

    9)   PROS/QPOE
    Class:   MODERATE and FATAL - help files give solutions
    Workaround:     YES
    Symptoms:       YES - fatal errors
    
    The tasks timsort, qpcopy, qpsort still have trouble with memory.
    If these tasks run and seem to take an extremely long time (>10
    minutes for 100000 photons) or give strange error messages ("can"t
    delete file"), try using a smaller sortsize buffer.
    Also, they may die at the end by having too many temporary files, or
    being unable to index.  This requires optimizing parameter settings.
    

    10)   PROS/QPOE on VAX
    Class:  MODERATE and FATAL
    Workaround:   YES
    Symtoms:      YES - 0 events in QPOE
    
    The conversion of FITS files to QPOE files on the VAX does not work
    all the time with the default qp_pagesize and qp_bucketlen parameters.
    We have worked around this problem by setting them to 512 and 1024
    respectively.  A user with an extra large FITS file may run into
    a limit with the number of temp files allowed in making the QPOE. 
    This error will be fatal.