Continuing my pass through the "bug fixes" section of the CommitFest,
I came upon this patch, which is contested. Here is my attempt to
summarize where things stand. As I understand it:
- Tom wants to revert to the previous behavior of accepting arbitrary
garbage, so that \d slkgjskld.jgdsjhgjklsdhg.saklasgh.foo.bar means \d
foo.bar.
- I want \d mydb.foo.bar to mean \d foo.bar if the dbname is mydb and
report an error otherwise; anything with dots>2 is also an error in my
view.
- Peter Geoghegan agrees with Tom.
- Stephen Frost agrees with me.
- Vik Fearing also agrees with me.
- Justin Pryzby, who originally discovered the problem, prefers the
same behavior that I prefer long-term, but thinks Tom's behavior is
better than doing nothing.
- Mark Dilger, Isaac Moreland, Garick Hamlin, Alvaro Herrera, and
Julien Rouhaud have commented on the thread but have not endorsed
either of these dueling proposals.
By my count, that's probably a vote of 4-2 in view of the preferred
solution, but it depends on whether you could Justin's vote as +1 for
my preferred solution or maybe +0.75 or +0.50 or something. At any
rate, it's close.
If anyone else would like to take a position, please do so in the next
few days. If there are no more votes, I'm going to proceed with trying
to fix up Mark's patch implementing my preferred solution and getting
it committed.
Thanks,
...Robert