roy@alanine.phri.nyu.edu (Roy Smith) (08/27/90)
We have discovered a minor problem with the software currently on
goober.phri.nyu.edu for processing the b.m.g.u postings. The current
algorithm for merging a new complete distribution with the previous
distribution plus accumulated updates involves throwing away any locus in
the accumulated updates if there is an entry with the same locus name in the
new distribution. No attempt is made to parse the dates in the loci to make
sure the locus on the tape is newer than the update one.
Imagine the following scenario. An entry is made to the data base
sometime before release N is frozen. The tapes for release N are cut, and
then an error is discovered in the locus and a new, corrected entry is made
in the database with the same locus name and ascession number (with the
particular entry that caused us to realize this, it was just changing a
"submitted to ..." reference to a real citation, but it could just as easily
have been something more serious). That revised entry reaches us via usenet
(before the tape gets here by US Mail) and gets put into our update database.
Then the tape shows up and gets merged into the database. The merging
program sees an entry on the tape with the same locus name and ascession
number as one in our updates database and tosses the update, leaving only the
older, incorrect, entry.
I consider this bug to be interesting, but not critical, so we've
just put it on our "to do" list. To the best of my knowledge, this only is a
problem with the Unix version of the software; I don't know if the VMS
version parses dates or not.
--
Roy Smith, Public Health Research Institute
455 First Avenue, New York, NY 10016
roy@alanine.phri.nyu.edu -OR- {att,cmcl2,rutgers,hombre}!phri!roy
"Arcane? Did you say arcane? It wouldn't be Unix if it wasn't arcane!"smith@mcclb0.med.nyu.edu (09/05/90)
In article <1990Aug26.231806.4043@phri.nyu.edu>, roy@alanine.phri.nyu.edu (Roy Smith) writes: > We have discovered a minor problem with the software currently on > goober.phri.nyu.edu for processing the b.m.g.u postings. The current > algorithm for merging a new complete distribution with the previous > distribution plus accumulated updates involves throwing away any locus in > the accumulated updates if there is an entry with the same locus name in the > new distribution. No attempt is made to parse the dates in the loci to make > sure the locus on the tape is newer than the update one. The VMS software uses a drop-dead date to decide whether an item in the UPDATE bank supercedes the tape release version or not. Bugs excepted then, the VMS stuff should retain the new(er) UPDATE sequence posting. +---------------------------------------------------------------------------+ |Ross Smith, Cell Biology, NYU Medical Center, 550 First Ave., NYC, 10016| |Phone: (212) 340-5356: FAX: (212) 340-8139 (Alternate NYUMC) (212) 340-7190| |E-Mail: SMITH@NYUMED.BITNET (BITNET), SMITH@MCCLB0.MED.NYU.EDU (Internet)| +---------------------------------------------------------------------------+