WIP: Merge release/42.2 into main #2264
Conversation
…e, PGline, PGlseg, PGpath, PGpoint, PGpolygon, and PGmoney 1) PGobject.setValue(null) can be used to set the wrapped value to null Note: previously the sub-classes of PGobject threw NPE for setValue(null) 2) Return null from `getValue` in case the typed null is received from the database 3) Assume PGBinaryObject#lengthInBytes == 0 means the value should be set as null when sending data to the database see pgjdbc#1870
Removal of LibPQFactory annotations was not done right
…s 42.2.14 and earlier enum[] => String[] jsonb[] => String[] json[] => PGobject[] fixes pgjdbc#1876
* fix: avoid removal type annotations on "this" so the source archive is buildable "this" type annotations are Java 8+, so we no longer need to remove them. * fix: PgTokenizer was ignoring last empty token (pgjdbc#1882) * fix: PgTokenizer was ignoring last empty token fixes pgjdbc#1881 * fix: handle smallserial correctly fixes pgjdbc#1897 (pgjdbc#1899) * feat: add smallserial metadata (pgjdbc#899) PostgreSQL 9.2 adds a SMALLSERIAL data type, this reports the correct metadata information when a column is a smallserial (int2 with sequence), similar to how a serial or bigserial data types are reported. * fix:remove osgi from karaf fixes Issue pgjdbc#1891 (pgjdbc#1902) * Change default of gssEncMode to ALLOW. PostgreSQL can deal with PREFER but there are cloud providers that did not implement the protocol properly. Using PREFER seems to cause more problems than it solves Co-authored-by: Vladimir Sitnikov <sitnikov.vladimir@gmail.com> Co-authored-by: Jorge Solorzano <jorsol@gmail.com>
* fix: avoid removal type annotations on "this" so the source archive is buildable "this" type annotations are Java 8+, so we no longer need to remove them. * fix: PgTokenizer was ignoring last empty token (pgjdbc#1882) * feat: add smallserial metadata (pgjdbc#899) PostgreSQL 9.2 adds a SMALLSERIAL data type, this reports the correct metadata information when a column is a smallserial (int2 with sequence), similar to how a serial or bigserial data types are reported. * fix:remove osgi from karaf fixes Issue pgjdbc#1891 (pgjdbc#1902) * Change default of gssEncMode to ALLOW. PostgreSQL can deal with PREFER but there are cloud providers that did not implement the protocol properly * update CHANGELOG and posts for 42.2.17 * fix up CHANGELOG * Update README, gssEncMode for release Co-authored-by: Vladimir Sitnikov <sitnikov.vladimir@gmail.com> Co-authored-by: Jorge Solorzano <jorsol@gmail.com>
javadoc can be built with mvn -Pjavadoc ...
This dependency is no longer needed for package build
…are not enabled by default pgjdbc#1637 made it impossible to activate binary transfer for the types except the default ones.
…e version Due to issue checkstyle/checkstyle#5711 some violations in the Javadoc were not reported. This PR is to pass validation with the new Checkstyle version.
…sion Due to issue checkstyle/checkstyle#5711 some violations in the Javadoc were not reported. This PR is to pass validation with the new Checkstyle version.
…arses pgjdbc classes (PR pgjdbc#2010)
…ties Previously TestUtil always overwrited gssEncMode.
…ad client certificates
Java 11 does not drain input stream completely during the handshake,
so alert("bad client certificate") is not received until pgjdbc sends
the startup message, which is too late.
So we allow a generic CONNECTION_UNABLE_TO_CONNECT code for "bad client certificate"
case.
Typical usages are changeSocket(createSocket()), so changeSocket should not be called from createSocket().
jorsol
force-pushed
the
master-merge
branch
4 times, most recently
from
29d983a
to
28bef6d
Compare
|
Ok, now it should be easier to review the changes, I have merged some fixes for the documentation that was on release/42.2 but not in main to facilitate the process. Now comparing the changes should be straightforward, I cross my fingers to not have messed up with some conflicts as having two parallel branches really makes things harder to maintain, in the future, all the changes should go to the main branch first, and only then, make a backport to an older branch. To recap, this PR merges all the changes from |
|
OK, so can you break up these changes into different PR's and we can start pushing them into main? I'd say make the docs one change. Then logically break up the changes as they occurred in the PR's on release/42.2. Thanks very much for doing this! |
Sure, I will break up this into different PR's as time allows me, there is a commit for geometric types that could also be a different PR and make this more manageable. |
jorsol
force-pushed
the
master-merge
branch
3 times, most recently
from
52e60d6
to
0170286
Compare
|
Update: the documentation, readme, and changelog are now updated on main, also cherry-picked the fix to avoid null in geometric types, probably Now it looks like this PR is more manageable and could give an overview of changes in release/42.2 that are not in main, probably I will have to redo this PR as there are some unintended changes, and double-check the conflicts but I see some progress. |
jorsol
force-pushed
the
master-merge
branch
3 times, most recently
from
1a3c73a
to
361da49
Compare
Signed-off-by: Jorge Solórzano <jorsol@gmail.com>
|
sure, no problem, |
|
This looks great. I scanned through a diff of this as release/42.2 and I don't see anything missing. |
|
I have to take another run through the changelog, but then I'd like to release it. |
jorsol
force-pushed
the
master-merge
branch
2 times, most recently
from
2a4b7ef
to
c15e553
Compare
|
Ok, so far it looks like the changes that remain are mostly cosmetic, closing this now as everything should be in place, as usual, it's probably a good idea to test what is in main and that there are no major regressions or issues. |
|
@jorsol Thanks so much for doing this. It really helped! |
This is just a test merging release/42.2 into main to see differences.