# bug log started: SRE, Sun Apr 28 15:27:41 2002
#
ID              sq1
STATUS          closed
DESCRIPTION     sfetch can't follow paths out of cwd if using SSI
NOTED_DATE      6 Mar 02
NOTED_XREF      -
REPORTED_BY     Zhirong Bao (bao@genetics)
REPRODUCE_WITH  Testsuite/bug-1-sfetch-paths
REPRODUCE_ON    any
CLOSED_DATE     6 Mar 02
CLOSED_XREF     -
CLOSED_ACTION   -
//
ID              sq2
STATUS          closed
DESCRIPTION     Fails to compile on Alpha. strtoull unresolved.
NOTED_DATE      SRE, Sun Apr 28 15:29:55 2002
NOTED_XREF      STL6 p.41
REPORTED_BY     Sam Griffiths-Jones (sgj@sanger.ac.uk)
REPRODUCE_WITH  ./configure; make
REPRODUCE_ON    waterbug (Alpha, Digital Unix) 
CLOSED_DATE     28 Apr 02
CLOSED_XREF     STL6 p.41
CLOSED_ACTION   -
//
ID              sq3
TITLE           weight segfaults, at end of alignment output 
STATUS          CLOSED
XREF            -
REPORTED_BY     Robin Dowell
CLOSED_DATE     SRE, Fri Feb 20 08:36:09 2004
DESCRIPTION     
   weight segfaulted on a_bacterial.stk:
       % valgrind weight -f .8 a_bacterial.stk
   shows a fault in MSAFree(). new->sqlen is being double free'd.
   fix: set to NULL after free'ing it in msa.c:MSASmallerAlignment().
//
ID              sq4
TITLE           shuffle -0 and -1 have limited sequence length range
STATUS          CLOSED
XREF            -
REPORTED_BY     Mike Pheasant (m.pheasant@imb.uq.edu.au)
OPENED_DATE     SRE, Wed Jun 30 10:31:45 2004
CLOSED_DATE     SRE, Wed Jun 30 10:31:46 2004
DESCRIPTION     
   StrMarkov0() and StrMarkov1() were counting in floats. This
   limits length range to ~6 digits. Changes so it counts in 
   integer, then normalizes to floating point probabilities; 
   this gives a range of ~2GB sequence length (assuming 32-bit int).
//
ID              sq5
TITLE           FChoose(), DChoose() return impossible residues
STATUS          CLOSED
XREF            -
REPORTED_BY     Mike Pheasant (m.pheasant@imb.uq.edu.au)
OPENED_DATE     SRE, Thu Jul  1 11:40:10 2004
CLOSED_DATE     SRE, Thu Jul  1 11:40:11 2004
DESCRIPTION     
   Another problem w/ shuffle -0 and shuffle -1 was seen; they
   could generate residues that had never occurred in the original
   sequence. Problem is in FChoose(),DChoose(), where the 
   "bulletproofing" against machine float representation was to
   choose 1 of N randomly, in the rare case that roll >= total_sum.
   To fix, force the bulletproofing to choose a nonzero p[i].
//
ID              sq6
TITLE           Phylip output should only put - for gap chars.
STATUS          CLOSED
XREF            -
REPORTED_BY     Arno Meiler <meiler@o2inline.de>, Wed Jul 28 2004
OPENED_DATE     SRE, Wed Jul 28 19:18:05 2004
CLOSED_DATE     SRE, Wed Jul 28 19:18:06 2004
DESCRIPTION     
   As of at least PHYLIP 3.6, . characters are not permitted as gaps,
   and Joe documents this. Rewrote the phylip parser, and changed
   the phylip example file in Formats.
//
ID              sq7
TITLE           seqsplit appends extra garbage characters
STATUS          CLOSED
XREF            -
REPORTED_BY     Tom Jones <tajones@genetics.wustl.edu>, Wed Aug 4 2004
OPENED_DATE     SRE, Wed Aug  4 20:09:32 2004
CLOSED_DATE     SRE, Wed Aug  4 20:09:33 2004
DESCRIPTION     
   malloc of (length) instead of (length+1) for seqfrag.
   seqsplit may append garbage chars as a result, under many
   circumstances.
//




