Thanks for feedback. I imagined this feature could be useful in a couple of contexts, though mainly in the sense of documenting query optimization efforts.
In one case you may be tweaking cost parameters and having a built in record of the parameters in the explain output can make that more reliable.
In a support context, it can help communicate all the details behind the selection of a query plan, perhaps on IRC support.
David J. --
I originally conceived the notion for the feature while off in the deep end thinking about automatic cost parameter selection.
It seems plausible that the ideal theoretical costs would be influenced by server activity (io/cpu utilization) and various buffer states at a point in time. This kind of optimization work is a little beyond my understanding and capability so this was a first step in thinking about this topic. It turned into a trivial user text output feature, but implementing output formats that are more machine friendly still makes a lot of sense.
Best Regards,
-Robert
On Wed, Nov 13, 2013 at 4:16 PM, David Johnston <polobo@yahoo.com> wrote:
Stephen Frost wrote > * Robert Berry (
> berrydigital@
> ) wrote: >> This is my first attempt at writing a patch, so it's pretty simple. > > Neat! > >> seq_page: 1.000000, rnd_page: 4.000000, cpu_tup: 0.010000, >> cpu_ind: 0.005000, cpu_op: 0.002500 >> amenabled: 11111111111 >> >> The bit vector is enable variables in the order listed in cost.h, >> though mainly provides a high level view on whether or not any strategies >> are disabled. > > While I like the general idea, I have to admit that I don't particularly > like the format and I'm not sure why it makes sense to have this as part > of 'explain'? Why not do a 'show all;' ahead of the explain?
I kinda get the theory behind this but, WRT formatting, explain can output multiple formats and any patch affecting said output should provide for changing all of them. Having each of the sample outputs in the post would allow for comments from those who would not generally apply such patches.
