| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
The usage of, the unusual, `&` in a filename caused troubles as the filenames in the file strings in the location element were not translated to XML.
|
|\
| |
| | |
Spelling corrections for src directory
|
| |
| |
| |
| |
| |
| |
| |
| | |
Spelling corrections as found by codespell and in #561.
Some reported problems were already fixed, others are fixed here, with some exceptions (a,o.):
- "referenceby" in defgen.cpp as this is in the output and I cannot oversee the consequences (looks like none, but ...)
- "HANGEUL_CHARSET" left as is as in some MS documentation is written: 'HANGUL_CHARSET: Also spelled "Hangeul". Specifies the Hangul Korean character set.' (https://docs.microsoft.com/en-us/openspecs/windows_protocols/ms-wmf/0d0b32ac-a836-4bd2-a112-b6000a1b4fc9).
|
|/ |
|
| |
|
|\
| |
| |
| | |
https://github.com/albert-github/doxygen into albert-github-feature/bug_endblock_msg
|
| |
| |
| |
| | |
Consistency
|
|\ \
| | |
| | |
| | | |
https://github.com/albert-github/doxygen into albert-github-feature/bug_gitversion
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The original version has as features:
- getting the git version number for usage in doxygen
- making the doxygen version number inclusion dependent on the VERSION file
The disadvantage of the chosen methodology was that an extra correction step was necessary, by defining getter methods to retrieve the values this correction can be hidden.
The information is coming from different sources:
- the VERSION file
- the git "repository and build system (when present)
Furthermore there are a couple of places where the version information is used (a.o. doxygen and doxywizard executable, though the doxygenwizard was only done "half hearted")
The handling of the VERSION file has been made in such a way that it is comparable with the generation of the git version changes.
For a better abstraction the version handling is all done in a separate directory.
|
| |/
|/| |
|
| | |
|
|/ |
|
|\
| |
| | |
Correcting errors in xml generation and xsd schema
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
- `xmlgen.cpp` make count test identical to restrictions in `generateXMLForMember`
- `index.xsd` add missing selection possibility
- `compound.xsd`
- add missing `block` possibility for htmlonly
- add missing `private` and `protected` possibilities
- get order in description type right
- make `sect[1-4]` more flexible
- adding missing elements like `msc`, `linebreak` at the appropriate places
- adding `vhdl` coloring enumeration values
- adding `docbook` possibility for images
results have been tested (a.o.) doxygen tests, doxygen examples, doxygen docu, CGAL source code.
|
|/
|
|
| |
the `declname` and `defname` should also be converted (compare as well the routine `generateXMLForMember`)
|
| |
|
| |
|
|
|
|
| |
in the XML output
|
| |
|
|
|
|
|
|
| |
Caused the test (079) to fail with a SIGSEGV, but larger projects exit
with a success return code and the generated XML is either truncated or
empty. Weird.
|
| |
|
|\
| |
| |
| | |
https://github.com/mosra/doxygen into mosra-xml-namespace-members-in-file-scope
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
For better consistency with the HTML output, where each file
documentation lists (and links to) all members of given namespace. This
also makes it possible to be consistent with the HTML output in case
a namespace is not documented and thus all its member detailed docs
should be put into corresponding file docs instead.
In order to be backwards compatible and avoid breaking stuff for
existing users of the XML output, this is controlled by a new
XML_NAMESPACE_MEMBERS_IN_FILE_SCOPE configuration option that defaults
to NO.
Note that this, unlike the HTML output, will put the whole detailed docs
into the file scope instead of just listing them. It's up to the user of
the XML output to deduplicate this information. It can be done for
example by comparing member ID prefixes with compound ID -- íf
different, the detailed docs are already somewhere else.
|
| |
| |
| |
| | |
Added a Slice-optimized output mode.
|
|\ \
| | |
| | | |
update to the experimental sqlite3 generator
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
In order for sqlite3gen to re-use parts of the XML generator, XMLCodeGenerator
needs to be declared in a header file. I parroted how other generators handle
this and have it working in both the XML and sqlite3 outputs, but I don't have
any sense of whether this is otherwise "right".
|
| | | |
|
|/ /
| |
| |
| |
| |
| | |
2017 debug)
The not initialized variable `l` caused and endless loop in the VS2017 debug version, variable should not have been used.
|
| | |
|
| |
| |
| |
| |
| |
| |
| | |
table of contents
- add the in page table of contents for XML
- add the possibility to have maximum levels in the in page table of contents (possible per output type and per in page table of contents. Default is show all)
|
|/
|
|
| |
Found via `codespell`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a C++11 `enum class` was present in an anonymous namespace (usually
in *.cpp files), the XML output was emitting warnings similar to the
following:
Internal inconsistency: member False does not belong to any container!
And the XML output was rendering bogus IDs for enum values starting with
`dummy_`, such as:
dummy_1a96ab6574751fdf6a53ceec8a3896c45daf8320b26d30ab433c5a54546d21f414c
The fix is to call memberOutputFileBase() on the enumeration itself and
not on the enum value, that way it provides correct file base that
corresponds to file base of the enumeration. There's also a new test
that checks this.
Note: this assumes that enum values belong to the same compound as enums
themselves. In my experience that was always the case and there's no
broken test after this change, so I hope I didn't break anything.
|
|\
| |
| | |
Full support for strong and typed enums in the XML output
|
| |
| |
| |
| |
| |
| |
| | |
There's a new boolean `strong` attribute on the `<memberdef>` element
that defines whether the enum is a classic or a strong one.
Tests were updated to verify both cases.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The <type> element (used for function return type and variable type) is
now reused also for underlying enum type.
C++03-style enum types that don't have the underlying type specified
have the <type> tag as well, but empty to indicate that the underlying
type is unspecified.
Added a new test case for this.
|
|\ \
| | |
| | | |
Provide template parameters also for type aliases in the XML output
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Until now, probably due to C++03 assumptions, typedefs and type
aliases skipped printing of the template parameter specifications to
the XML output. That's now fixed, so the following type alias will
properly contain <templateparamlist> in the XML output:
template<class T> using Vec = std::vector<T>;
Added also a test case that verifies this.
|
|\ \
| | |
| | | |
Expose TOC placeholder in XML output
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Currently the XML output for a page was generated the same
independently of whether the \tableofcontents command was present in
the page source or not. Because of that, the users of the XML output
had no chance of knowing whether given page should have a TOC or not.
There's now a new <tableofcontents> element that gets added in case
the TOC was requested. As it is trivial to populate the TOC on user
side by simply enumerating the <sectN> elements, the element is empty
and acts just as a boolean.
|
|\ \ \
| |/ /
| | | |
Provide page brief in <briefdescription> of XML output
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Until now the brief description of pages was prepended to the
<detaileddescription> element and the <briefdescription> element was
not present at all, which meant there was no easy way to extract the
brief description for purposes of creating a page index, for example.
With this patch, the brief description is included in both
<briefdescription> and <detaileddescription>, thus duplicated, to
avoid backwards compatibility issues.
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The <sp> element in <programlisting> has a new optional attribute
`value`, which contains value of given invalid ASCII character. In
case of space, the `value` attribute is omitted.
Use cases for this: including snippets of *very esoteric* languages,
markup that makes use of the advanced ASCII formatting characters or
for example highlighting a console output containing ANSI color codes
(which is my case, in fact).
Regarding backwards compatibility -- as files with such ASCII
characters are very rare, I don't expect this minor difference in the
output to be a problem. Besides that, such ASCII characters are often
replaced by a space in many applications anyway.
A test snippet was extended to contain a special character so this
difference in behavior could be verified.
|
| |
|
|
|
|
| |
Protected against NULL pointer of variable al
|
|
|
|
| |
Documentation section
|
|
|
|
|
| |
Modified the memberdef table so that it includes all elements xmlgen
dups for a memberdef.
|
|
|
|
| |
Value of <float> in bitfield was not escaped when writing to XML
|
|
|
|
| |
improve performance
|
|
|
|
|
|
|
| |
- made static build option for all windows targets
- made wizard flex interactive
- add 64bit support for libiconv on windows build
- updated test mechanism
|
| |
|