Hi Nikolay, I did a quick look at yout patch and have some questions/points to discuss. I like the idea of the patch and think that enum reloptions can be useful. Especially for some frequently checked values, as it was mentioned before.
There are few typos in comments, like 'validateing'. I have two questions about naming of variables/structures: 1) variable opt_enum in parse_one_reloption named in different way other similar variables named (without underscore). 2) enum gist_option_buffering_numeric_values/gist_option_buffering_value_numbers. Firstly, it has two names. Secondly, can we name it gist_option_buffering, why not? As you mentioned in previous mail, you prefer to keep enum and relopt_enum_element_definition array in the same .h file. I'm not sure, but I think it is done to keep everything related to enum in one place to avoid inconsistency in case of changes in some part (e.g. change of enum without change of array). On the other hand, array content created without array creation itself in .h file. Can we move actual array creation into same .h file? What is the point to separate array content definition and array definition? -- Aleksandr Parfenov Postgres Professional: http://www.postgrespro.com Russian Postgres Company
