octave-forge-io-2.6.4-bp155.1.9

- Update to version 2.6.4:
  * Bug fixes:
  - fromJSON.m, toJSON.m: some improvements for Java-based BIST.
  - make saving .ods/.xlsx files work again with OCT interface.
  - fix opening .xlsx files made using Google docs (bug #60830).
  - allow other file extensions than .xls/.xlsx when saving to a
    different file.
  - fix writing text strings in .ods files (bug #60825).
  - improve file extension processing.
  * Code improvements
  - toJSON.m, fromJSON.m: made much more robust.
  * Deprecated function:
  - object2json (superseded by toJSON).

- Update to version 2.6.3:
  * Bug fixes:
  - xlsopen.m: don't wipe user selected reqinterface upon first
    call to xlsopen after loading io package (bug #59273, bug
    [#59277])
  - Improve handling of POI interface, esp. OOXML support (bug
    [#59273])
  - Make reading .ods using OCT interface more reliable (bug
    [#59273])
  * Code improvements:
  - Spreadsheet I/O interfaces echoed to screen when invoking
    test scripts with "verbose flag.
  - Significant speed improvements when reading .xlsx files with
    OCT interface, thanks to Dennis Zeilstra (bug #59277)
  * New functions
  - toJSON and fromJSON: convert Octave objects into a JSON
    string and vice versa, contributed by Ketan M. Patel (patch
    [#9980]).

- Update to version 2.6.2:
  * Bug fixes:
    + Fix Named range read errors for .ods files with OCT interface.
    + dbfread.m: open files in little-endian mode (fix by Rafael Laboissière).
    + xlsopen.m: improve filename extension checking (.xlsm, .xlsb and .sxc
    were not recognized).
    + xlsclose (private/__COM_spsh_close__.m): add missing output formats for
    COM interface (esp. .ods; also .wk3 and .wk4).
    + xlsread.m: implement options argument (it *was* mentioned in the help
    but didn't work)
    + Improve character encoding for COM interface (bug #59203); credits to
    Markus Mützel
  * Code improvements:
    + xlsfinfo.m: implement 'verbose' flag to echo used interface info.
    + io_testscript.m: minimize delay, improve texinfo.
    + csv2cell's maximum line length has been increased to 32768 characters
    (bug #58618).
    + Remove autotools (bootstrap, configure). Avoids potential problems and
    makes code easier to maintain. Kudos to Kai Torben Ohlhus (bug 357081).
    + Reintroduce xmlread and xmlwrite functions.
    NOTE: xmlread may provoke Octave crashes on some Linux systems (see
    bug #58004)

- Update to version 2.6.1:
  * io-2.6.1 is meant to polish a few prominent bugs introduced unfortunately
    in the coarse of code overhaul for io-2.6.0.
  * Bug fixes:
  - xlsopen: reinstate default OCT interface (reason that on systems w/o
    Java or with non-matching Java bit width "no spreadsheet I/O support
    was found" for .ods and .xlsx).
  - ods*.m wrapper functions: properly implement output args (bug #58045).
  - Wrong concatenation of indices when reading .xlsx with OCT interface.
  * Still no fix for the XML functions :-(
- Update to version 2.6.0:
  * Bug fixes:
  - Fixed some concealed bugs in getxmlattv.m & __OXS_spsh2oct__.m.
  - dbfwrite.m: made more robust as regards heterogeneous data columns, just
    skip them rather than abort.
    Also fixed writing just one data row (bug #57669)
  - read_namelist.m: fixed a minor typo.
  - OCT interface: properly return negative values formatted as "(<value>)"
    (bug #57812).
  - The XML functions xmlread.m and xmlwrite.m have been (temporarily)
    removed (see bug #58004). Once that bug is fixed these functions will be
    reinstated.
  * Code improvements:
  - Most user-visible spreadsheet I/O functions have been adapted such that:
  * All supported spreadsheet formats / file types can be read and written
    with the xls*.m functions. So it isn't required anymore to read
    .xlsx/.xls with xlsread, and .ods with odsread, etc.
  * The ods*.m functions still exist but now are mere wrappers for the
    corresponding xls*.m functions. They are (softly) deprecated but will
    be retained for a while (to at least somewhere in 2022 ?).
    As they are mere wrappers they have all the functionality of the
    corresponding xls*.m and previous ods*.m functions.
    This has been done to cut down on code maintenance; it wasn't really
    manageable to keep two separate function sets in sync that behind the
    scenes provided close to 95 % overlapping internal functionality. The
    choice for keeping xls*.m rather than ods*.m functions was made for
    Matlab compatibility - that has no ods*.m functions.
  - The xlsread.m help text now contains an extensive explanation of
    spreadsheet I/O interfaces.
  - The message indicating which interfaces were found is now silenced but
    can be shown by adding an argument "verbose" to xlsread / xlsopen /
    xlswrite. See the help for xlsread, xlsopen and xlswrite.
  * Other changes
    The previously separate READ-ODS.html and READ-XLS.html docs in the doc/
    subdirectory have been replaced by a new combined html background document
    rewritten from scratch: Spreadsheet-IO-in-Octave.html

- Update to version 2.4.13:
  * Bug fixes:
  - COM interface: restore writing formulas to spreadsheet
  - OCT interface: ignore NA and NaN values when writing .ods and
    .gnumeric (bug #56375)
  - OCT interface: don't truncate arrays when using single
    cell-range input with oct2ods and oct2xls functions (bug #56444)
  - csv2cell: properly process leading tilde in file name (bug #53591)
  * Code improvements:
  - COM interface: better support for creating file formats other than
    .xlsx
  - ditto for UNO interface
  - Remove references to legacy Java package, as io depends on Octave
    >= 4.0 where core Octave should have built-in Java support anyway
  - Replace calls to strmatch() with calls to str(n)cmp(i)()
  - Replace calls to __octave_config_info__() and __have_feature__() by
    javachk() (bug #47480)
  * Various:
  - The io package now needs Octave >= 4.0.0, due to dependence on javachk

- Update to version 2.4.8:
  * csv2cell(): Improve ignoring CR (char (13)) in csv files (bugs
    [#50979] and #50297)
  * OCT interface:
  - More robust identification of worksheets (spreadsheets
    written with POI interface were misinterpreted)
  - Overhauled reading .xlsx files; catches more types:
    inlineString, Boolean, ...
  * read_namelist.m: improve reading when ending char is not fist
    on line (bug #52242)
  * Replace utf82unicode with native2unicode v.v. in Octave 4.4+
  * If no filename extension is given, default to .xlsx rather
    than .xls
  * xlsread.m and odsread.m now accept a function handle, an
    output options struct and interface as arguments after the
    third (range) argument, in any order.
  * all spreadsheet I/O functions now accept an empty string as
    worksheet name, implying the first worksheet.

- Update to version 2.4.5
  * New functions.
  * Bug fixes.

- Update to version 2.4.3
  * io_xls_testscript.m, io_ods_testscript.m: return results rather
    than bail out when hitting errors
  * cell2csv: increase precision hen writing double values
  * Bug fixes

- Update to version 2.4.2
  * Changed several calls to (now deprecated) octave_config_info into
    calls to __have_feature__.m
  * Add support for complex numbers in read_namelist.m and write_namelist.m
  * Bug fixes

- Update to version 2.4.1
  - xmlread.m and xmlwrite.m as replacement of the undocumented,
    unsupported and unusable binary code.
  - New function tidyxml.m:  simply strips characters < 32 or > 126 from
    character arrays or cell arrays (including mixed cell arrays).
  - test_spsh.m now shows a nice summary of test results for each interface.

- Update to version 2.2.11
  * Bug fixes

- Update to version 2.2.10
  * Support for Named Ranges
  * Bug fixes

- Update to version 2.2.9
  * Bug fixes

- Update to version 2.2.8
  * Bug fixes

- Update to version 2.2.7
  * Bug fixes

- Split from octave-forge package, version 2.2.6

- Update to version 2.4.12:
  * Bug fixes:
  - dbfwrite.m: avoid crash and vectorize writing data to file
    (bug #53899)
  - dbfwrite.m: add unwind_protect block to be able to avoid
    dangling file handles in case of errors; better checks on
    data column homogeneity; texinfo (help) header fixes
  - A little more robust Java check
  - csv2cell: improper string indexing (bug #54407), fix by
    Orion Paplowski
  * Code improvements:
  - POI interface: adapt to POI-4.0.0 (new getCelltype enums)
  - POI: overhaul cell type handling when reading and writing
  - dbfread.m: improve speed when reading a subset of data

- Update to version 2.4.11:
  * OCT interface: fix reading shared formula results (bug #52875)
  * OCT interface: fix a rare case of nested text cell XML with
    special formatting attributes (bug # 53298)
  * OCT interface: don't over-allocate memory when reading text
    cells in xlsx (bug 533910, credits to Peter Giles
  * OCT interface: return correct date values for .ods file (bug
    [#53401])
  * dbfwrite.m: don't prepend space before string values
  * bug #53510: fix detection of Java version for Java > 8
  * OCT interface: (part of) bug #53459, i.e. reading .xlsx files
    with a more complex mix of chartsheets, macrosheets and
    worksheets, and exploring such files with xlsfinfo. Writing to
    such files however isn't guaranteed to be bug free. However,
    other interfaces (POI, UNO) provide a workaround.
  * COM interface: clearly distinguish between worksheets and e.g.,
    chartsheets when referencing sheets by number.