Refactor decoding arrays

[bokken]

Refactor decoding arrays to make more extensible and to more closely align the binary and String processing.

[codecov-io]

Codecov Report

Merging #1194 into master will increase coverage by 0.02%.
The diff coverage is 83.33%.

@@             Coverage Diff              @@
##             master    #1194      +/-   ##
============================================
+ Coverage     70.02%   70.05%   +0.02%     
  Complexity     4160     4160              
============================================
  Files           188      188              
  Lines         17537    17550      +13     
  Branches       2872     2874       +2     
============================================
+ Hits          12281    12294      +13     
+ Misses         3977     3976       -1     
- Partials       1279     1280       +1

[bokken]

@vlsi or @davecramer
Can you help me interpret the failing travis issue:

+wget --quiet -O /tmp/openj9.tar.gz https://github.com/AdoptOpenJDK/openjdk9-openj9-releases/releases/download/jdk-9%2B181/OpenJDK9-OPENJ9_x64_Linux_jdk-9.181.tar.gz
The command "./.travis/travis_install_openj9.sh" failed and exited with 8 during .
Your build has been stopped.

It does not seem like this is code related, but I am not familiar enough with travis to know what exiting with 8 means.

[vlsi]

That means wget exit code was 8, and it means there was a server error.
It turns out the URL is no longer valid.

We might want to use https://ci.adoptopenjdk.net/job/openjdk9_openj9_build_x86-64_linux/ (or even OpenJDK10-Open9J), however filename is different for each build there.

[jorsol]

This makes me think that the openj9 is not production-ready and that it actually does not provide any benefit in testing, after all it's the same openjdk with a different engine, and in fact it slows down the testing... I will send a PR drooping that job for consideration.

[bokken]

@vlsi and @davecramer, is there a page in the documentation covering what data types are supported for binary transfer? This PR adds support for binary transfer of OID_ARRAY and it is not clear to me if that should be documented somewhere.

[davecramer]

@bokken Good question. I suspect not, but I would like to see it documented somewhere

[bokken]

Can we target this for 42.3?

[davecramer]

I was just talking to @vlsi about that and no we don't care about 9 anymore. I have a PR to try 14 #1844

[vlsi]

@bokken , there's a default qualifier for NonNull

pgjdbc/pgjdbc/src/main/java/org/postgresql/package-info.java

Lines 6 to 9 in 5801e32

	@DefaultQualifier(value = NonNull.class, locations = TypeUseLocation.FIELD)
	@DefaultQualifier(value = NonNull.class, locations = TypeUseLocation.PARAMETER)
	@DefaultQualifier(value = NonNull.class, locations = TypeUseLocation.RETURN)
	package org.postgresql;

, so please refrain from adding @NonNull in trivial cases like arr and connection below

    public void populateFromString(@NonNull A arr, List<@Nullable String> strings, @NonNull BaseConnection connection) throws SQLException {

[vlsi]

Have you tried ArrayDecoder<A extends @NonNull Object> here? (I assume you intend that A should be non-nullable)

https://checkerframework.org/manual/#inherited-wildcard-annotations

[davecramer]

@bokken is this ready to merge ?

[bokken]

@davecramer Yes!!!
Thanks to @vlsi getting the checker framework happy, this should be good to go.