Brar Piening
Brar Piening
Since I definitely want #4181 to be solved for v7 and I think, @roji's [suggestion](https://github.com/npgsql/npgsql/issues/4181#issuecomment-1003536437) would probably work, I've implemented a test for how I think load balancing should work...
This adds a souce generator that parses the pg_type.dat and pg_range.dat (copies from their src/include/catalog/ folder) files, that PostgreSQL itself uses to initialize a cluster, to generate a NpgsqlDatabaseInfo. Closes...
This was broken by changes introduced with the multiple hosts feature.
As discussed in https://github.com/npgsql/npgsql/pull/4461#issuecomment-1127537959 and below we could allow connections with an empty connection string if all the information needed to connect a server is available from environment variables.
With the current implementation of load balancing between multiple servers (connection string setting "Load Balance Hosts=true") we move on to the next host without incrementing our internal round robin counter...
Closes #3387 ~~If I read the code correctly it is as simple as that.~~ Narrator: "It wasn't." ~~I clearly must be missing something.~~ Narrator: "Yes"
This is the first of possibly multiple PR's that evolve from my attempt to first understand, then potentially clean up and lastly document internal interfaces that look scary, overly wide...
As already mentioned in https://github.com/npgsql/npgsql/issues/922 PostgreSQL 9.5 has changed the default for the `include_realm` setting to `1` (`true`), meaning that it is the default setting for all currently supported PostgreSQL...
In https://github.com/nunit/nunit/pull/2895#issuecomment-398906513 @ChrisMaddock wrote: > I surprised so many have sneaked in. 🙁 I guess ideally, we’d run our tests in a couple of different cultures. This might be a...
This is a prove of concept for allowing empty connection strings. It is not meant as actual PR to be merged but rather as a means to gather feedback and...