==================================
PDFlib 10.0.1 (September 22, 2022)
==================================

- 2022-09-22 (bug #7325)
  Imported listbox fields without an appearance stream were rendered without
  taking the /TI entry in the form field dictionary into account (which
  corresponds to the "topindex" option of PDF_create_field()).

- 2022-09-22 (feature #3009)
  Added support for PHP 8.2.

- 2022-09-22 (bug #7341)
  The callout line of imported FreeText annotations was not placed correctly
  and ended up at a wrong location on the output page if the imported page
  contained a /Rotate key with a value different from 0.

- 2022-09-21 (bug #7340)
  When importing Line annotations with PDI the caption position could be wrong.

- 2022-09-21 (bug #7294)
  PostScript Type 1 fonts are not supported as host fonts on macOS. However,
  they no longer trigger an exception but only a warning since they can still
  be used as file-based fonts.

- 2022-09-15 (feature #3016)
  Updated to curl 7.85.0.

- 2022-09-15 (bug #7330)
  The PDFlib .NET binding did not work with current AWS Linux Lambda Docker
  images.

- 2022-09-15 (bug #7313)
  The PHP binding was not working on RedHat Linux 8 for ARM 64.
  
- 2022-09-14 (bug #7332)
  Changed the target moniker framework (TFM) entry in the .NET samples and
  documentation to "<TargetFramework>net6.0</TargetFramework>".
 
- 2022-09-06 (bug #7326)
  With option "script=_auto" for Textline and Textflow output OpenType
  features were applied in an unexpected way with certain fonts, which could
  also thwart shaping.

- 2022-08-21 (feature #2945)
  Added correct table checksums to TrueType/OpenType fonts retrieved as
  hostfonts on macOS.


=================================
PDFlib 10.0.0p6 (August 23, 2022)
=================================

- 2022-08-23 (bug #6915)
  Reduced the memory usage for optimizing output documents.

- 2022-08-22 (bug #7268)
  Improved the performance for SVG color fonts when SVG documents are shared
  for the rendering of multiple glyphs.

- 2022-08-17 (bug #7168)
  In Unicode-based bindings a wrong error message was produced if an
  error message parameter was not valid UTF-8.

- 2022-08-12 (bug #7128)
  The default "_none" for the "script" option had caused unpredictable
  application of OpenType features with certain fonts. The default value of the
  script option was now changed to "latn".
  
- 2022-08-10 (bug #7299)
  Fixed a problem with Textflow that caused leading spaces in justified or
  left-aligned text after implicit line breaks.

- 2022-08-08 (bug #7306)
  The fact that PDF_load_font() options "subsetlimit" and "autosubsetting"
  are deprecated and no longer functional since PDFlib 10 had not been
  properly documented. These options are still accepted but have no effect.

- 2022-08-05 (bug #7307)
  Trying to produce a backslash in an option value in front of
  the closing brace as documented with two backslashes caused an unexpected
  exception.

- 2022-08-01 (bug #7308)
  Textflow save/restore did not work as expected for the following options:
  adjustmethod, alignment, hyphenchar, tabalignchar, avoidemptybegin,
  avoidbreak, minlinecount, minspacing, maxspacing, hortabsize, hortabmethod,
  ruler, tabalignment, parindent, lastalignment, leftindent, tf_rightindent,
  spreadlimit, nofitlimit, shrinklimit, fixedleading, mark, leader, shadow.


===============================
PDFlib 10.0.0p5 (July 27, 2022)
===============================

- 2022-07-27 (feature #2995)
  Significantly improved PDI performance for reading huge arrays with object
  references as they may appear in malformed PDF documents with tags or layers.

- 2022-07-27 (bug #7312)
  In the C++ wrapper avoid pollution of the global C++ namespace with the
  "PDF" name.

- 2022-07-27 (feature #3012)
  Added support for Ruby 3.1.

- 2022-07-25 (bug #7303)
  Using option list delimiter characters in the text for a FreeText
  annotation caused the unexpected exception "Internal error:
  assertion failed in file 'p_annotap.c', line 951".

- 2022-07-25 (bug #7300)
  Updated to OpenSSL 1.1.1q.

- 2022-07-21 (feature #3003)
  Updated to curl 7.84.0. As a consequence networking support is no longer
  available on System Z.

- 2022-07-21 (bug #7157)
  Fixed an inconsistency in TextFlow that trailing spaces were preserved
  with centered and right-aligned text, while they were discarded with
  left-aligned text. Now trailing spaces are preserved in all three cases.
  This is relevant if trailing spaces are visible, e.g. by underlining or
  when using matchboxes with visual appearances.

- 2022-07-11 (feature #2985)
  Updated to zlib 1.2.12.
  
- 2022-07-10 (feature #3004)
  Implemented PDI optimizations for reading direct number and string objects
  where the type and value are retrieved. Also, the dictionary vs. stream
  check is eliminated to speed up processing of entities which are surely
  not dictionaries such as some layer and Tagged PDF objects.

- 2022-07-07 (bug #7301)
  The thread-safe PDFlib library for PHP 7.3 and 7.4 on ARM64 has actually
  been built in non-threadsafe mode.

- 2022-07-03 (bug #7296)
  Significantly mproved PDI performance for importing documents with layers
  and tags by eliminating redundant collection of data for logging.

- 2022-07-03 (feature #3001)
  Significantly improved PDI performance for reading huge dictionaries as they
  may appear in malformed PDF documents with layers or Tagged PDF properties
  and rolemaps.

- 2022-06-30 (bug #7289)
  Fixed the problem that on some platforms the Python binding did only work
  with specific Python versions.

- 2022-06-30 (bug #7291)
  Improved the performance of importing Tagged PDF.

- 2022-06-26 (bug #7292)
  Significantly improved PDI performance for reading huge dictionaries with
  object references as they may appear in malformed PDF documents with layers. 

- 2022-06-23 (bug #7288)
  Fixed a crash that occurred when using tagging options in table functions
  when no tagged PDF output was actually created.

- 2022-06-23 (bug #7290)
  Added support for Ruby 2.7 on FreeBSD.

- 2022-06-23 (bug #7283)
  Linker errors because of duplicate symbols could occur when linking an
  application against PDFlib and the ICU library.


===============================
PDFlib 10.0.0p4 (June 14, 2022)
===============================

- 2022-06-14 (bug #7275)
  Linker errors occurred when linking a Windows application against PDFlib
  and the curl library because of duplicate symbols.
  
- 2022-06-09 (bug #7278)
  The implementation of the SVG CSS property "font-weight" suffered from two
  problems:
  - The "font-weight" property was ignored unless also a "font-style" property
    was present.
  - The keywords "bold" and "normal" were not recognized.

- 2022-06-01 (bug #7261)
  Importing the layer structure of documents with an unreasonable amount of
  layers could take very long. Now PDI ignores layers if the number of
  layers in the document exceeds 250.

- 2022-05-30 (bug #7269)
  Placing an SVG graphics inside a template where a <text> element contained
  only the Unicode character U+FEFF ZERO WIDTH NO-BREAK SPACE caused an
  unexpected exception "fit_textline(blind): Function must not be called in
  'template' scope".


==============================
PDFlib 10.0.0ME-p3 (May 19, 2022)
==============================

- 2022-05-19 (bug #7263)
  Using PDF_begin_document() option "labels" while not also specifying option
  "groups" resulted in the unexpected exception "Option 'labels' requires
  suboption 'group' if used with this function".
  
- 2022-05-17 (bug #7266)
  The parser for SVG glyph definitions in OpenType SVG color fonts made an
  unwarranted assumption about the structure of SVG glyph definitions
  containing multiple glyphs. When this assumption was violated, an unexpected
  exception "Internal error: assertion failed in file 'svg_parse.c', ..."
  occurred.


==============================
PDFlib 10.0.0p3 (May 13, 2022)
==============================

- 2022-05-13 (bug #7257)
  Invalid PDF/UA output could be created with weakly structured PDF/UA output
  when violating the header hierarchy, which was not rejected as required.
  
- 2022-05-13 (bug #7258)
  Fixed a memory leak that could occur when an exception was thrown during
  import of annotations.

- 2022-05-13 (bug #7264)
  The keywords "annotations" and "widgets" for option "taborder" of
  PDF_begin_page_ext() and PDF_end_page_ext() did not work as expected for
  PDF 2.0 output.

- 2022-05-11 (bug #7259)
  Fixed a performance problem in pCOS and PDI triggered by /Root dictionaries
  containing a large number of direct objects.

- 2022-05-06 (bug #7256)
  U+200D ZERO WIDTH JOINER was replaced with a space glyph if the font did
  not contain a glyph for U+200D. Now the glyph is dropped in this case
  without a replacement.

- 2022-05-06 (bug #7249)
  Invalid PDF output was produced when using certain rare OpenType CFF CID
  fonts with subsetting where the Private DICT data was not stored adjacent
  to the Font DICT INDEX data. So far this was only observed with the
  Noto Serif Japanese and Noto Sans Japanese fonts by Google that were produced
  in 2015. Newer versions of these fonts do not trigger the problem.

- 2022-05-05 (bug #7255)
  Opening an image in info mode without closing the corresponding image
  handle before a subsequent call to PDF_close_pdi_document() caused the
  unexpected exception "Internal error: assertion failed in file 'p_pdi.c',
  line 1941".
  
- 2022-05-05 (bug #7250)
  Invalid PDF/UA output was created when importing pages from multiple
  PDF/UA import documents where the heading levels on subsequent pages from
  different documents violated the PDF/UA heading level requirements. Now this
  problem is detected and an error message is produced.

- 2022-04-28 (bug #7248)
  Improved the error message when calling PDF_open_pdi_page() after opening
  an encrypted document with PDF_open_pdi_document() with option
  "requiredmode=minimum" and/or option "infomode=true" while not specifying
  a password. The error message now says explicitly that this is not possible
  in info mode.

- 2022-04-28 (bug #7246)
  Using the 32-bit .NET Universal package on Windows in Debug mode resulted
  in the .NET runtime error message "A call to PInvoke function
  'PDFlib_dotnet!PDFlib_dotnet.PDFlib+PInvoke::PDF_new_csharp' has unbalanced
  the stack.

- 2022-04-28 (bug #7242)
  PDF_open_pdi_page() in PDF/UA mode didn't detect some structure problems
  related to the heading hierarchy on the imported page vs. headings in the
  output document:
  
  - Only the top element of the imported hierarchy was tested against the
    existing header hierarchy, which could lead to incorrect acceptance of
    the page.
    
  - The check for consistent heading levels was wrong if the output document
    had been closed and more pages were imported into another output document
    (e.g. when splitting a document page by page), which could lead to
    incorrect acceptance of the page.

  Both problems could result in PDF/UA validation errors of the generated
  output document.

- 2022-04-25 (bug #7241)
  Importing a page containing a link twice from a PDF/UA-1 document into
  a PDF/UA-1 output document resulted in invalid PDF/UA-1 output.

- 2022-04-19 (bug #7243)
  The Perl and Ruby bindings on macOS were not built as Universal macOS
  Binaries.

- 2022-04-01 (bug #7238)
  When XMP document metadata was created and multiple document information
  dictionary entries were created with non-ASCII keys, it could happen that
  not all document information dictionary entries were mirrored in XMP.

- 2022-03-28 (feature #2972)
  Added several CJK fonts with builtin metrics which can be used without
  external font files provided "embedding=false" is supplied.

- 2022-03-17 (bug #7236)
  The PDFlib 10.0.0p2 package for MVS was broken.


================================
PDFlib 10.0.0p2 (March 17, 2022)
================================

- 2022-03-17 (bug #7219)
  The .NET Universal package was not working with the 32-bit .NET runtime.

- 2022-03-17 (bug #7225)
  A call to PDF_create_gstate() with log level "warning=3" triggered an
  unjustified warning "Function must not be called in 'page' scope".

- 2022-03-17 (bug #7233)
  Restored support for Perl 5.26.

- 2022-03-15 (bug #7232)
  Linking with PDFlib on MVS failed because of unresolved symbols.

- 2022-03-10 (bug #7228)
  Calling PDF_info_textflow() with the "used" keyword after PDF_fit_textflow()
  had been called another time after "_stop" had been returned by the preceding
  call to PDF_fit_textflow() could cause an out-of-bounds read and potentially
  a crash.

- 2022-03-09 (bug #7227)
  Caused by an incorrect mapping it could happen that the glyph U+1E9E LATIN
  CAPITAL LETTER SHARP S was displayed when glyph U+00DF LATIN SMALL LETTER
  SHARP S was requested if a glyph for U+1E9E was available in the font.

- 2022-03-07 (bug #7224)
  When the SVG "kerning" property contained an explicit length value the SVG
  import still applied auto-kerning from font kerning tables if available. Now
  the correct behavior is implemented that the "kerning" length value modifies
  the inter-glyph spacing while font kerning tables are ignored.

- 2022-03-04 (bug #7218)
  Importing a tagged PDF document while creating tagged PDF output could
  cause the unexpected exception "Internal error: assertion failed in file
  'pc_output.c', line 3208".

- 2022-03-04 (bug #7214)
  When importing PDF documents with large tagged PDF data structures while
  creating PDF/A-1 output it could happen that the PDF/A-1 limit of a maximum
  of 8192 array elements was exceeded, resulting in invalid PDF/A-1 output.

- 2022-03-04 (bug #7223)
  For a font with kerning information it could happen that kerning was applied
  although option "kerning=false" was set, for example with justified text
  alignment or when option "xadvancelist" was specified. Also for SVG import
  option "kerning=false" means now that font-based kerning according to
  SVG property "kerning=auto" is not applied.

- 2022-02-24 (bug #7216)
  An empty /Next array in an action dictionary caused an unexpected
  exception "Internal error: assertion failed in file 'pc_output.c', line 2843"
  when importing the corresponding page with annotations.
  

===================================
PDFlib 10.0.0p1 (February 15, 2022)
===================================

- 2022-02-15 (bug #7199)
  When form fields were imported from a PDF page that was rotated via the
  page dictionary's /Rotate key, the orientation of the resulting form fields
  in the output was incorrect.

- 2022-02-15 (bug #7209)
  Using fit method "fit" or "slice" with PDF/UA output caused the unexpected
  error message "Images and graphics must be tagged as 'Artifact', 'Figure' or
  'Formula' in PDF/UA".
  
- 2022-02-15 (bug #7195)
  When importing form fields from a document that contained the same form
  field multiple times, this triggered the error message "name already exists
  in the document". Now the field is imported once and subsequent instances
  of the same field are ignored with a warning.

- 2022-02-15 (bug #7206)
  If an image contained a clipping path, a user-specified clipping path
  specified via suboption "clipping" of PDF_fit_image() option "matchbox"
  was ignored. Now both clipping paths are applied.
 
- 2022-02-10 (bug #7202)
  Creating too many tagged items of structure type "TOCI" could result in
  an unexpected error message or a crash.

- 2022-02-09 (bug #7200)
  The Windows 64-bit package for PHP 8.1 x64 TS contained the 32-bit DLL
  instead of the 64-bit variant.

- 2022-02-07 (bug #7197)
  The attempt to fill a form field that could not be imported resulted in the
  unexpected error message "Parameter 'type' is empty". Now the fill attempt
  is ignored with a warning.

- 2022-02-04 (bug #7196)
  Importing invalid check box form fields with inconsistent field
  characteristics flags caused an error message "Font not specified for field
  'C:N < 25'".

- 2022-02-04 (bug #7194)
  For source builds the OpenSSL sources are no longer distributed separately.
  Now they are part of the PDFlib source package.

- 2022-02-04 (bug #7191)
  Importing invalid form fields with loops in the field hierarchy caused
  the error message "Parameter 'type' is empty".

- 2022-02-04 (bug #7189)
  The PDFlib 10.0.0 .NET NuGet package contained an incorrect version
  number in the package description.

- 2022-02-04 (bug #7187)
  Importing an annotation with a color value outside of the allowed range
  between 0 and 1 caused an error message "Floating point parameter '...' has
  bad value ..." in PDF_end_page_ext(). Now the color values are clamped to
  the allowed range.

- 2022-02-04 (bug #7186)
  pCOS generates an artificial name for broken form fields that lack a
  required /T entry. The attempt to fill such a field caused a crash.

- 2022-02-04 (bug #7181)
  When a name of an imported form field contained a '#' character, the
  attempt to fill that field could fail with the error message
  "Imported form field '...' does not exist (required for filling)".

- 2022-02-02 (bug #7084)
  Drop illegal trailing null bytes in the names of imported JavaScript and
  other entries in the /Names tree to avoid creation of U+FFFD REPLACEMENT
  CHARACTER in the output.

- 2022-01-31 (bug #7142)
  OpenType color fonts: kerning pairs involving the space glyph were missing
  in the generated Type 3 font. 
  
- 2022-01-28 (bug #7179)
  Creating an empty Type 3 font caused an unexpected exception.


================================
PDFlib 10.0.0 (January 25, 2022)
================================

- Implements a variety of new features and options, see PDFlib Tutorial
  and API Reference for details.

- Removes a number of long deprecated API methods and options; see
  PDFlib Migration Guide for details.