| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
init_priority attribute
* we cannot use weak symbols in MinGW, so we avoid init_priority for symbol
initialization by compiling the empty definitions into
sciteco-minimal but the real ones into sciteco
(had to add new file symbols-minimal.cpp)
* this fixes compilation/linking on LLVM Clang AND Dragonegg
since their init_priority attribute is broken!
this will likely be fixed in the near future but broken versions
will be around for some time
|
|
since including glib.h on LLVM-Clang (32-bit) results in NULL being
redefined to 0 and compiler warnings being emitted when NULL is used
as sentinels
|
|
* flow control and other structures have not been documented this ways.
I have not yet decided whether they should be documented in separate
sections or use the documentation tool.
|
|
* storage size should always be 64 (gint64) to aid macro portability
* however, for performance reasons users compiling from source might
explicitly compile with 32 bit integers
|
|
* also encapsulates data properly (previously there were many public
attributes to avoid permission issues)
* new class also cares about saving/and restoring scroll state.
now, buffer/q-reg edits and temporary accesses do not reset
the scroll state anymore.
|
|
* there was a dependency on interface initialization.
it did not cause issues because destruction order was
by chance.
* introduced INIT_PRIO and PRIO_* macros to easy initialization order declaration
(using a PRIO_* formula makes code self-documenting)
* also used this to clean up QRegisterTable initialization
(we do not need the explicit initialize() method)
* also used to clean up symbols initialization
|
|
memory mapping the entire file has been benchmarked to be less
efficient than the old implementation (because of more than
doubling page faults).
A lengthy comment has been written to discuss different
implementations of file reading.
|
|
* file must be in primary memory for scintilla
* we cannot write to scintilla's buffer memory directly
* so mapping the file is best: in the best case it is not copied to primary memory and resides in kernel cache
* in any case, mapping to memory is faster than read()ing into primary memory
* copying from mapped virtual memory to scintilla buffer (via SCI_APPENDTEXT) is faster
than copying from primary memory
|
|
* works with/without save-points (i.e. in batch and interactive mode, both were broken)
* improved file-saving performance (avoid buffer gap removal)
* only root can preserve the ownership of a file owned by another user after file saving
|
|
|
|
|
|
|
|
|
|
|
