dateutils-0.4.9-bp154.1.34

- Update to 4.9:
  * allow to specify epochs in nanoseconds, %s%N (#123)
  * fix nanosecond arithmetic (#124)
  * make cross-compilations easier (#128)
  * mention default timezone (#129)
- Drop upstreamed patches:
  * 0144b15dda4b1ea4fdbeca9f3ecb128eb98772fd.patch
  * a0ebd0037df973aed14779b51d59da3edc506b6a.patch

- Add 0144b15dda4b1ea4fdbeca9f3ecb128eb98772fd.patch and
  a0ebd0037df973aed14779b51d59da3edc506b6a.patch: fix test suite
  run on 32bit archs.

- add upstream signing key
- simplify spec file

- Update to 0.4.8:
  * be symmetric about differences, A - B := -(B - A) for A < B
  * fix security issue when reading TZMAP_DIR from environment
  * fix printing of negative 0 (-0)
  * fix issue with tzdata transitions in files with leap second
  * adjustments that are now cut off at expiration time (#121)
- Drop dateutils-remove-deprecated-header.patch: not needed anymore

- Add dateutils-remove-deprecated-header.patch: Remove deprecated
  glibc sysctl.h header.

- Update to 0.4.7:
  * strptime mimics return code behaviour of dateconv
  * use unsigned arithmetic for leap second conversion
  * fix up ymds before adding days or business days
  * drain prchunk if filled with many small lines

- Update to 0.4.6:
  * underspecified ISO weekday dates can be parsed
  * adding days to underspecified dates results in proper dates
  * all tools with -S|--sed-mode now support -E|--empty-mode to
    output an empty line when a line cannot be read.
  * dateround supports day, week, month and year classes now
    (issues/92)
  * datediff can determine difference in nanoseconds between
    two time values (issues/94)

- Simplify conditions
- Run spec-cleaner

- Update to 0.4.5:
  * dgrep accepts %c as qualifier in expressions (issue 81)
  * tests work regardless of wall clock time (issue 84)

- Ensure neutrality of descriptions.
- Trim idempotent %if..%endif guards.

- Update to 0.4.4:
  Incompatible changes:
  * suffix `m` is no longer accepted as a synonym for `mo`
    with date-only input, issue #76
  Bugfixes:
  * expose BSD routines (fgetln()) to yuck
  * for dates passed to dateround(1) that coincidentally
    match the roundspecs do read them as dates
  * wrong timestamps read via -i %s signal error
  * facilitate bmake build
  * time rounding on date-only input keeps the date unchanged
  * dateseq's short-cut iterator (2 date arguments) does not
    interfere with the 3-argument version
  * adding 0 date or time units does not change the summand
  * datezone on times (without date) will return times
  * zones are singletons now, opened and closed only once

- Add upstream patches 35041f4d.patch and 841c635b.patch: Do not
  use Singapore for testing previous transitions, but rather
  Jakarta. Fixes build with timezone 2022g.

- Update to 4.10:
  * do not resort to daisy sequences when incomplete dates are present (#131)
  * upon end-of-buffer finalise the last line (#141)
  * let --empty-mode|-E and --sed-mode|-S play nicely together (#139)
  * adapt CLI behaviour of strptime(1) to the other tools (#142)

- bump version 0.4.3
  * Features:
  - base expansion works for times now
  * Bugfixes:
  - durations in months weeks and days are calculated
    like durations in months and days, consistency
  - am and pm indicators in inputs are handled properly
  - military midnights decay when not printed in full

- revert non-portable changes of the last commit (SLE 11 build)

- bump version 0.4.2
  * Features:
  - allow %- specifiers to turn off padding (as GNU date does)
  - support matlab day numbers, as mdn or matlab
  * Bugfixes:
  - fix build issue on FBSD 11
  - allow zones to transition at INT_MAX (mapped to never)
- lint spec file
- move from bitbucket to github link
- remove patch which is upstream now:
  0001-fix-when-printing-zones-transitioning-at-INT_MAX-in-.patch

- fix incompatibility with tzdata/tzcode >= 2017a,
  0001-fix-when-printing-zones-transitioning-at-INT_MAX-in-.patch

- bump version 0.4.1, this is a bugfix release.
  * Bugfixes:
  - bug/50, keep end-of-term (ultimo) property in dateseq(1)
  - allow today/now for ywd dates in 1 parameter dateseq(1)
  - allow different calendars for start and end date in dateseq(1)
  - round time in dateround(1) in zone then convert --from-zone
  - allow SPC padded numerals in %d input format
  * See info page examples and/or README.
- remove patch which is upstream now:
  0001-Revert-chore-be-more-portable-in-sxe-matlab.m4.patch

- fix .mex file extension for octave plugin,
  0001-Revert-chore-be-more-portable-in-sxe-matlab.m4.patch
- explicit octave plugin file list
- remove *.la files

- bump version 0.4.0, this is a feature release.
  * Bugfixes:
  - document %g/%G format specifiers
  - bug/42, accept NUL characters in input
  - bug/45, correctly display Gregorian and ISO week dates in one
    line
  - bug/46, adhere to ISO group's official formatting of week
    dates
  - bug/47, allow rounding of Epoch based timestamps
  - bug/48, avoid crash for void input to strptime(3)
  * Features:
  - datetest supports --isvalid to conditionalise on date/time
    parsing
  - all tools support --from-locale to parse localised input
  - tools with output formatting support --locale to format
    output according to locale
  * See info page examples and/or README.

- bump version 0.3.5, this is a bugfix release.
  * Bugfixes:
  - bug/40, distinguish between numerals-only dates and durations
  - tests will succeed independent of current date
  * See info page examples and/or README.
- remove dateutils-0.3.4-fix-test-126.patch (upstream applied)

- Fix build:
  * add dateutils-0.3.4-fix-test-126.patch (from upstream)
    (bnc#966644)

- remove old obsoletes line for devel package

- bump version 0.3.4, this is a bugfix release.
  * Bugfixes:
  - bug/39, MacOSX endianness detection
  - bug/38, ddiff day-only durations on date/times
  - dadd +2m bug is fixed, regression
  * See info page examples and/or README.
- remove patches (upstream applied):
  * 0001-fix-build-for-compilers-without-anonymous-struct-ini.patch
  * 0002-fix-dateadd-unix-timestamp-bug.patch

- Correct previous changelog entries (incorrect patch names)

- add 0002-fix-dateadd-unix-timestamp-bug.patch

- bump version 0.3.3, this is a feature release.
  * Features:
  - to clarify purpose and avoid name clashes prefix binaries
    with "date". This results in: dateadd, dateconv, ... etc.
  - provide compatibilty through configure switch --with-old-names
  - provide single digit years through %_y
  - allow rounding of ISO-week dates (ywd) to week numbers
  * Bugfixes:
  - dashes behind a date do not count as tz indicator
  - UTC/TAI/GPS special coordinated zones work on systems without
    leap second support
  * See info page examples and/or README
- add 0001-fix-build-for-compilers-without-anonymous-struct-ini.patch
  to support broken compilers (SLE 11, RHEL 6, etc.)

- bump version 0.3.2, this is a bugfix release.
  * Bugfixes:
  - out of range minutes will be discarded
  - bug 30 (malicious input crashes dconv) has been fixed
  * Features:
  - military midnight stamps are supported (T24:00:00)
  - 8601 ordinal dates (year + doy) are recognised directly
    (`-f yd`)
  - strptime(1) can behave in a locale-dependent way
  * See info page examples and/or README.

- sort out distros with and without octave

- bump version 0.3.1, this is a bugfix release.
  * Bugfixes:
  - octave/matlab code is distributed fully
  - negative durations with refined units are minus-signed only
    once
  - ddiff is entirely anticommutative now
  - tests don't fail if zones don't exist on the build system
  - dseq with empty ranges will no longer produce output (just as
    seq(1))
  - arbitrary integers are not interpreted as time anymore
  - when converting from zone info properly clear zone difference
    for %Z
  - dseq(1) will automatically resort to +1mo and +1y iterators
    for wildcarded ymd dates
  - dadd(1)'ing ywd dates with output as ymd works properly now
  * Features:
  - ddiff can output nanosecond diffs
  - automatic fix-up of dates is documented now
  - parser errors and fix ups are reported through return code 2
  - dseq with no `-f|--format` stays in the calendric system of
    the start value instead of converting all output to ymd
  * See info page examples and/or README.

- bump version 0.3.0, this is a feature release.
  * Features:
  - dgrep supports time zones both for the expression and the
    input
  - timezones can be specified by alternative codes and
    [tzmap files][1]
  - new tool dzone to inspect date/times in multiple timezones in
    bulk
  - new tool dsort to sort input chronologically
  - gengetopt and help2man maintainer dependencies removed
  - lilian/julian inputs via `-i ldn` and `-i jdn`
  - ymcw dates now follow ISO 8601 in using 07 to denote Sunday
  * Bugfixes:
  - ddiff takes differences between a unix epoch stamp and a
    date/time
  - zone converter assigns correct sign to zone difference when
    using %Z
  - weekdays are properly calculated from epoch stamps (issue 24)
  * See info page examples and/or README.
    [1]: http://www.fresse.org/dateutils/tzmaps.html
- enable Octave/Matlab plugin
- remove useless devel package

- bump version 0.2.7, this is a feature release.
  * Features:
  - dgrep supports -v|--invert-match like grep
  - output specifier %G is supported for compatibility with POSIX
  - ddiff calculates year-day differences
  - ddiff calculates ISO-week date differences
  - ddiff output can be zero and space padded through 0 and SPC
    modifier
  - zoneinfo database on AIX >= 6.1 is taken into account
  * Bug fixes:
  - ddiff can calculate full year differences, [issue 21][1] fixed
  - dseq now accepts %W, %V output formats, [issue 22][2] fixed
  - builds with clang >= 3.3 work again, [clang bug 18028][3]
  * See info page examples and/or README.
    [1]: https://github.com/hroptatyr/dateutils/issues/21
    [2]: https://github.com/hroptatyr/dateutils/issues/22
    [3]: http://llvm.org/bugs/show_bug.cgi?id=18028

- bump version 0.2.6, this is a bug fix release.
  * issue 19, -q|--quiet no longer sends some commands into an
    inf-loop
  * netbsd test failures are fixed (due to missing leap seconds)
  * AIX builds are supported (getopt_long() is part of the code
    now)
  * internally the test harness is migrated to the cli-testing tool
    this fixes an issue when tests are run in directories with odd
    names (spaces, dollar signs, etc. in the path name)
  * See info page examples and/or README.