This release adds caching for the Describe protocol exchange, which is used to validate that declared parameter and column types are consistent with the schema. Because the schema is typically static at runtime, we can cache the results of these checks on a per-pool basis and save an extra round-trip during each Prepare exchange.

Pooled Session constructors now have extra parameters for cache-tuning, as described in the associated doc:

The pool maintains a cache of queries and commands that have been checked against the schema,
eliminating the need to check them more than once. If your program is changing the schema on
the fly than you probably don't want this behavior; you can disable it by setting the
commandCache and queryCache parameters to zero.

By default both caches have size 1024.

This release also changes SkunkException to include the full diagnostic message in getMessage rather than only in toString. This improves error reporting with MUnit.

CE3 version of 0.0.28

• removes Scala 3.0.0-RC2/3, adds Scala 3.0.0

Same as 0.0.26, sorry.

This is the CE3 version of v0.0.26.

This is the CE2 series. The equivalent CE3 release is v0.1.1.

• Dependency updates, including removing Scala 3.0.0-RC1 and adding 3.0.0-RC3.
• Removed Enumeratum support from enum codec for Scala 3 (it had been using .withDottyCompat)
• JSON module is now published for Scala 3, using Circe 0.14 milestones (Scala 2 version remains on Circe 0.13).

Equivalent to v0.0.25 but for CE3.

This release is for CE2. If you want CE3 the matching release is v0.1.0.

• Removes Scala 3.0.0-M3 and adds 3.0.0-RC2
• Bind argument tracing is now more readable.
• Add eimap operation for Codec (Niklas Klein).
• Doc updates. (Vova Polischuk)

Minor updates.

• Documentation improvements (Anton Kovalevsky, lambdista)
• Fix _float4 and _float8 codecs (Loránd Szakács)
• Hack up mdoc to show nicer toStrings.
• Improve handling of socket disconnects (see "Sessions" in the reference doc).
• Remove Scala 3.0.0-M2, Add RC1
• Update all dependencies

• Session parameters may now be customized, which is necessary in order to use Skunk with Redshift (Brandon Stilson).