Crash dumps generated with breakpad were not made use of to an extent
that would justify the extra maintenance overhead that it comes with. As
Windows builds have recently been broken by it, now is a good time to
retire it.
In the TypesWidget, right-click+Load New Types was opening the dialog
for editing an existing type instead of creating a new one. This would
either result in an error (for atomic types) or the old type being
deleted on success.
The QByteArray must be kept alive as long as its contens are used. Also
in this function, Core()->renameFunction() is not used to avoid sending
multiple signals for a single edit action.
When triggered through a keyboard shortcut, the dialogs shown here would
be positioned at the last position of the context menu or (0, 0), which
is not desired. Using a currently shown widget as the parent fixes this.
This builds a real tarball, as opposed to the flawed GitHub-generated
one, and also includes the following changes:
Individual builds now have dedicated names like "linux-x86_64". The
structure in the yml is now very similar to how it is in rizin.
Since that means builds are renamed, the filename has also been changed
from the meaningless "ccpp.yml" to "ci.yml", as that would have happened
sooner or later anyway and now will not produce additional intermediate
rename states.
The workflow name inside that file is now also just "CI" since adding
"Cutter" there is redundant.
* Updated GraphGridLayout documentation
* Don't use potentially misleading name "segment tree" .
Not exactly segment tree (although sometimes called that), and for the purpose of high level understanding how graph layout works doesn't matter what it is. Any data structure which provides required queries could be used.
Co-authored-by: Kārlis Seņko <karlis3p70l1ij@gmail.com>
This fixes a double-free in getSignatureInfo() when the string was
already freed by the intermediate CutterJson. But actually just using
this CutterJson makes more sense than the string anyway, so let's do
that.
The old JsonModel rebuilt the entire structure, defeating the main
purpose of a custom model in comparison to the existing QTreeWidget.
And since we are holding the json structures ourselves anyway, such a
custom model does not have many benefits anyway.
Using Q_GLOBAL_STATIC meant that the CutterCore was destructed late as
part of a binary destructor. It would then free the RzCore, calling for
example the fini callbacks of all plugins. However global destructors in
shared library plugins may have already been run at this point, leading
to for example rz-ghidra's decompiler_mutex being used after
destruction.
Instead of the Q_GLOBAL_STATIC-managed global object, we are now
handling the lifetime of the CutterCore ourselves and only injecting its
instance to be accessed globally. This can also be a first step towards
making the core instance completely local.
