Dave,
Just a few comments regarding your pg_dump lock timeout patch (in general I like the concept and agree with adding
it):
- No validity checking that the argument passed in has anything to dowith a number. The backend will do this, but it
strikesme as a bitodd to not do any checking at argument processing time.
- You call the argument 'wait time' in the documentation, but 'DELAY'in the command-line help. I'd recommend using
oneterm and stickingto it. You're already two lines in the command-line help, you canspell it out as 'WAIT_TIME' or
similar.
- getTables() uses different variables for each query, and I'minclined to agree with that approach to make following
thecodeeasier. I'd encourage you to add a new variable for thestatement_timeout query rather than reusing the lockqry
variable.Youcould even offset this by removing the unused delqry variable.
Otherwise, looks good to me.
Thanks,
Stephen
