querying on square-brackets, []

Dave Matthews matthews at greengenes.cit.cornell.edu
Tue Aug 13 09:49:39 EST 2002


This tace query takes 10 minutes to execute in GrainGenes,

  query find sequence [AL505438.1]

Whereas the following queries all complete in less than a second except
as noted.

  query find sequence (AL505438.1)
  query find sequence {AL505438.1}
  query find sequence "[AL505438.1]"
  query find probe [AL505438.1]      7 seconds
  find sequence [AL505438.1]

The Sequence class has 550,000 records, Probe has 17,000.
The object "[AL505438.1]" does exist in both classes but it doesn't matter,
same result with "[AL505438.2]" which doesn't.

Same result with ace4_9i,f,c,a and 4_7l.

The hardware is a Sun Ultra10 with 1GB RAM.  cachesize.wrm is set to 
CACHE1 = 32000
CACHE2 = 64000
so memory used by the process was ca. 350MB, which was available without
hitting the swap during these tests.

I realize that 'query find sequence [AL505438.1]' is not expected to return
the desired object but it's syntactically okay, right?  And there are
scenarios where a user could do it to me and tie up my sgifaceserver.  It
doesn't answer any other connections while this is happening, dead as a
stump.  (Actually it finally crashes and it's a bitch to restart, "Address
already in use".)

- Dave

---





More information about the Acedb mailing list