* use wstring in symbol and bookmark names
* fixed setting apppath in windows includes
* regard utf8 encoding in shared indexer commands
* regard utf8 encoding in shared indexer status
* used wstring in FilePath constructor where easily possible
* added FilePath concatenate method that accepts string parameter
* implemented using only wstring in MessageStatus
* used wstring in config
* changed order of compiler flags for CDB indexer commands so that manually added flags are appended to the flags form the compile command.
* removed usage of clang::tooling::FixedCompilationDatabase and used CompilationDatabaseSingle for empty and cdb indexer commands.
* removed reference to a specific clang tool (e.g. "cl" or "cl.exe") from compile command and prepended "clang-tool" before running the indexer.
* some parts of the indexer were called "manual" while others used the
name "empty". renamed everything that was called "manual" to "empty"
to conform to names in GUI
* removed post processing for anonymous typedefs. The storage transformation did not work in
cases where the typedef is recorded in a previously indexed header. No type information
was recorded while indexing the header a second time. But this type information would have
been required by the transformation.
* indexer processes wait if already too many storages queued
* storage fetcher waits if already too many storages queued
* increased size estimation for shared memory of indexer commands
* Added MessageFilterNewErrors to combine messages for new errors
bug id = 471
get rid of most clazy warnings (https://github.com/KDE/clazy)
* range loop use refs
* qcolor ctor with ints
* missing Q_OBJECT macro
* use .constfirst instead of .first
enabled ccache for unix systems for fasterr recompiling
* if ccache is in path we use it now
Before indexing the indexer commands are sorted according to byte size of the
indexed source file. Afterwards the list is split into two halves and each half
is shuffled. This ensures that bigger files tend to be indexed first, so that
no indexer thread runs idle at the end of indexing.
* moved storage related classes to data/storage folder
* implemented recording c++ unions as NODE_UNION instead of just using NODE_TYPE
* added transformation to merge anonymous classes/structs/enums/unions and typedef nodes
* added tests for this transformation
* added setting for enabling/disabling this transformation
* corrections in estimating size of intermediate indexer commands
* avoided clearing and refilling the indexed files list every time files get added
* changed memory growth policy of InterprocessIndexingStatusManager to double allocated memory when it gets full
* added potential to grow memory when adding crashed file paths
* disable windows error dialogs on application crash for indexer target
* show indexing dialog at the start of indexing task instead of waiting for processes to start
* fix escaping path to binary for indexer process
* join garbagecollector thread when stopping
* fix compile errors related to FilePath in deploy mode on windows
* TaskBuildIndex starts seperate indexer processes and communicates via shared memory
* indexer processes get restarted if they fail
* indexer processes are killed when the app closes or crashes
* added utility class SharedMemory to allocate and access shared memory, providing basic data structures
* added SharedMemoryGarbageCollector for keeping track of running instances and cleaning up shared memory in case of a crash
* show errors for source files that crashed during indexing
* added basic exception handling to task scheduling
* added option to turn off processes and use threads in preferences, but still use shared memory
fortune cookie message = Good news from someone dear is coming soon.
* mark all files with errors as incomplete
* fix incomplete label in codeview tooltip not updated properly
* added setting for canceling indexing for files with fatal errors
* avoid AST traversial for CXX translation units with fatal errors
* save all files within those translation units as incomplete
* show number of complete files in finished indexing dialog and overview stats
* display incomplete files hatched in graph view
* display incomplete file titles hatched in code view
* mention file incompleteness in tooltips
* removed derived Indexer classes and set ParserType via template parameter instead
* moved getSourceFileFromCDB from IndexerCxxCdb to IndexerCommandCxxCdb
bug id = 358
mlangkabel