We repeat what we did for the function handling:

- Deliberately and clearly output domain and d...

Re-jig the MemberTypeDef handling so that we deliberately and clearly
create a domain target and...

We pass the target handler to domain handlers as they are in a better
place to use it than the r...

Makes it easier to grep for and discover and is generally more explicit.

Probably others in the examples but this does no harm and was useful for
testing.

Thanks for helping with the Expat parser errors.

I don't like the idea of moving stuff to the base class for reuse
(composition/collaboration bei...

And use it in the class & function directives as well as in the index
processing.

We change the parser module code to raise different exceptions for the
IOError and ExpatError wh...

When some c++ source file is badly encoded, an ExpatError causes
breathe to crash. This patch pr...

As we have introduced new interfaces since the last version but haven't
broken anything (to my k...

We want to separate out all the ugly subprocess call and stuff into a
separate helper class to k...

To avoid confusion with the behaviour of path in the other directives.

So that we properly identify the usage of doxygenindex and
autodoxygenindex in the errors.

Covering the basics, though without a failing case as we need to chase
that up a bit.

Otherwise we're overloading the word and it is a bit confusing. We
should probably do the same f...

We add a config value to determine the directory that the doxygen xml is
generated to for the au...

We add a AutoDoxygenIndexDirective which takes the arguments and puts
out a DoxygenAutoNode into...

Instead of processing everything in place in the directive code, we
output a single DoxygenNode ...

To reflect the root directory from which to specify relative paths to
source files that we want ...

A simple custom node and transform inherited from the relevant classes
in docutils and registere...

Raised on the librelist. It is possible for a sectiondef element to have
a description which we ...

From libre list discussion.

I've left my try-except phase and have realised that there are more
graceful ways to handle thos...

Mistakenly left in there.

The code was getting confused when the .rst file that wanted the image
was not at the top level ...

And changed the documentation a little so that it isn't just an out of
context cut and paste of ...

We provide the very basics of support for the \image tag in doxygen
output by reading the name o...

And provide a helper method for getting paths relative to it.

Based on jo3w4rd's ticket.

As noted in the exception, if breathe_default_project had a value which
wasn't a key in the brea...

Found that properties were excluded by default.

Thanks for reporting the issue with programlistings.

Turns out that if we pass an extra empty string to the literal_block
node at the start then it r...

We have to parse the programlisting tags within para tags and then
provide renderers for program...

As reported by jo3w4rd on #65 on github.

Thank you for contribution handling asterisks in the embedded rst
comments.

We change the documentation to output a representation of the bad data
rather than feeding it th...

Comment styles in doxygen that use a leading asterisk, such as

/*!
* Commenting
* a bit
...

Only when we need it. We reschuffle the code so that we only look for
the default project if the...

To start semantic versioning.

Thank you for figuring out the update logic when doxygen xml files
change to get the rst files t...

We wrap up all the login in a FileStateCache object so that the internal
code only needs to know...

This means changes to doxygen xml files will caused rst documents that
use those files to be reb...

Thank you for contribution.

Thank you for contributions towards the overloaded function resolution.

We separate out the argument parsing and matching into separate function
for a bit more readabil...

With simple and more complex types.

In thanks for contributions towards the auto-directives which allow
on-the-fly generation of the...

Thank you for contributions towards the Python3 support.

Otherwise links to classes and other index level items fail.

The regression was introduced in:
...

So that __init__.py is syntax friendly for both python 2 & 3 as setup.py
needs to be able to imp...

Mostly personal preference.

By using the math node instead of the display math node when we detect
the start & end $'s which...

Involves picking up the "formula" xml nodes from the doxygen xml output
as children of the "para...

We use the same formatting code we have for the functions except we
check ahead as we don't want...

Previously we were outputting these section as a field list which looked
rather good but it adds...

Previously we were applying it if any "show" option had been provided at
all, here we make sure ...

We provide a ":show: header-file" option to the doxygenclass directive
which looks for the "incl...

... to squeeze our testsuite output through the latex pdf builder
without hitting the 6 level in...

Previously the CompoundDef renderer was choosing to indent its own
output but really it should b...

It hasn't been updated in ages and so was broken and expecting to use
all kinds of classes that ...

It is perhaps not flawed by itself but in conjunction with using
line_blocks to put the template...

In case it isn't immediately obvious.

We have an issue that we generally want to ignore most of the
pretty-printing whitespace inside ...

An issue was reported where a linked word inside a markup tag
(literal/fixed width font in this ...

For issue #46.

Forgotten previously.

Would be best to automate this somehow to look at the latest tag in the
g...

Thanks to LDiracDelta for pointing out the need for this.

For no really good reason, just personal preference.

This is a pretty random choice but we're slowing improving and it has
been a while since 0.6!

Thank you to hobu for the setup.py.

This implements a setup.py that was cribbed from sphinx's repository for breathe