postgresql18-orafce-4.16.2+git0.6a3cfa5-bp156.4.16.1

- Update to version 4.16.2+git0.6a3cfa5:
  * fix update script from 4.15-4.16 and prepare for 4.16.2

- Update to version 4.16.1+git0.3eaa566:
  * fix meson build + prepare for 4.16.1

- Update to version 4.16.0+git0.38f69ed:
  * More accurace calculating months_between when time part of input is not truncated.
  * fix build on pg19

- Update to version 4.14.6+git0.2b21fee:
  * prepare for 4.14.6
  * fix build pn pg19
  * remove useless usage of short life memory context
  * remove useless (and buggy) usage of short life memory context

- enable pg 18

- Update to version 4.14.5+git0.86c6e48:
  * prepare for 4.14.5
  * regression tests for fix oracle.replace_empty_strings and oracle.replace_null_strings
  * Fix in function oracle.replace_empty_strings detection of empty string of type character(n). Fix in function oracle.replace_null_strings generating empty string of type character(n).
  * the names of roles created by regress tests should to use prefix "regress_"
  * fix build on master (pg19)
  * fix formatting
  * fix formatting
  * post pgindent sync
  * pgindent formatting
  * typedef list for pgindent

- Update to version 4.14.4+git0.6d279e3:
  * prepare for 4.14.4
  * Fix pg_upgrade failure after updating orafce from version 3.13 or earlier to 3.14 or later.
  * fix white space
  * In PostgreSQL 18, the constant HAVE_LONG_INT_64 is removed. Instead the check SIZEOF_LONG == 8 should be used. HAVE_LONG_INT_64 was used to reduce useless comparation on some 32bit systems. Instead using check SIZEOF_LONG == 8, remove this condition completely and force check for int4 overflow everytime.

- Update to version 4.14.3+git0.f2afa97:
  * prepare for 4.14.3
  * fix broken offset for TupleDescAttr
  * update readme.meson_msvc
  * try to install additional package for compilation of orafce to CI

- Update to version 4.14.2+git0.eb0e492:
  * prepare for 4.14.2
  * fix crash when arguments of plvstr.rvrs specified substring out of passed string
  * fix check number of digits of umask value
  * remove security definer from dbms_alert.signal - it is not necessary since c40d56b
  * remove unsupported releases
  * fix build on old meson (tested 0.58.2)
  * enhancing meson build

- Stop building for EOL postgresql12.

- Update to version 4.14.1+git0.48e67e7:
  * prepare for 4.14.1
  * introduce new github workflows
  * Update README.asciidoc
  * fix regress tests on msvc
  * DiffUtils from gnuwin32 are necessary for ninja test on msvc
  * info about using meson
  * fix meson.build for msvc
  * support for meson

- Update to version 4.14.0+git0.f0ecbd3:
  * prepare for 4.14.0
  * prepare for 4.13.6
  * remove invalid alternative results of dbms_random test
  * fix dbms_random test for pg 12
  * fix error when orafce's triggers was executed on table with dropped columns
  * update year in info comment
  * fix white chars
  * Oracle does switch of low, high when low is higher than high arg of dbms_random.value. Do this in orafce too. Some cleaning in related regress tests. Attention: this change breaks alternative expected results.
  * regress test for new nvl function
  * fix Makefile and return type previously added nvl function
  * Add support to nvl(double precision, integer) and to_date(integer, text).

- Update to version 4.13.5+git0.ab95f82:
  * prepare for 4.13.5
  * export orafce_sys_extract_utc and orafce_sys_extract_utc_oracle_date for win build
  * With Oracle passing en empty string to the to_date(txt, fmt) function returns NULL whereas PostgreSQL returns '0001-01-01 BC'. This true whatever is the format passed as second argument. Change the behavior of function to_date(txt, fmt) to mimic Oracle.

- Update to version 4.13.4+git0.8f3f211:
  * prepare for 4.13.4
  * fix update script from 4.12 to 4.13
  * prepare for 4.13.3
  * umask is usually entered as octec number, then read it as octec number

- Update to version 4.13.4+git0.8f3f211:
  * prepare for 4.13.4
  * fix update script from 4.12 to 4.13

- Update to version 4.13.3+git0.0bf0f15:
  * prepare for 4.13.3
  * umask is usually entered as octec number, then read it as octec number

- enable pg17

- Update to version 4.13.2+git0.f8f8700:
  * prepare for 4.13.2
  * fix missing update of control file and regress tests

- Update to version 4.13.0+git3.25d74aa:
  * Fix description of GUC orafce.varchar2_null_safe_concat
  * Fix description of GUC orafce.varchar2_null_safe_concat
  * fix regress tests
  * Update README.asciidoc
  * Fix documentation formatting.
  * Add description of to_date(text,text) function and the error reported due to Oracle bug.
  * Update regression test result with error message change.
  * Raise an error when Oracle's to_date() bugs are reached. This concern dates before 1582-10-05 ('J2299159') usinf the 'J' format and dates before 1100-03-01 that cannot be verified due to a bug in Oracle.
  * remove obsolete sql signature of dbms_alert.defered_signal
  * update NEWS
  * Remove the BC part of the Oracle like Julian Day date
  * Fix wrong number of days add to julian day in to_char(text, text) function and add pg_catalog FQDN to the to_date() builtin function called. Add regression test.
  * remove obsolete file
  * Fix TO_DATE(text, text) function with the 'J' (Julian day) formatting. For Oracle this is the number of days since January 1, 4712 BC whereas for PostgreSQL this is the number of days since November 24, 4714 BC.

- disable pg11 builds as we will remove the EOL package soon