| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
* we were basing the glib allocators on throwing std::bad_alloc just like
the C++ operators. However, this always was unsafe since we were throwing
exceptions across plain-C frames (Glib).
Also, the memory vtable has been deprecated in Glib, resulting in
ugly warnings.
* Instead, we now let the C++ new/delete operators work like Glib
by basing them on g_malloc/g_slice.
This means they will assert and the application will terminate
abnormally in case of OOM. OOMs cannot be handled properly anyway, so it is
more important to have a good memory limiting mechanism.
* Memory limiting has been completely revised.
Instead of approximating undo stack sizes using virtual methods
(which is unprecise and comes with a performance penalty),
we now use a common base class SciTECO::Object to count the memory
required by all objects allocated within SciTECO.
This is less precise than using global replacement new/deletes
which would allow us to control allocations in all C++ code including
Scintilla, but they are only supported as of C++14 (GCC 5) and adding compile-time
checks would be cumbersome.
In any case, we're missing Glib allocations (esp. strings).
* As a platform-specific extension, on Linux/glibc we use mallinfo()
to count the exact memory usage of the process.
On Windows, we use GetProcessMemoryInfo() -- the latter implementation
is currently UNTESTED.
* We use g_malloc() for new/delete operators when there is
malloc_trim() since g_slice does not free heap chunks properly
(probably does its own mmap()ing), rendering malloc_trim() ineffective.
We've also benchmarked g_slice on Linux/glib (malloc_trim() shouldn't
be available elsewhere) and found that it brings no significant
performance benefit.
On all other platforms, we use g_slice since it is assumed
that it at least does not hurt.
The new g_slice based allocators should be tested on MSVCRT
since I assume that they bring a significant performance benefit
on Windows.
* Memory limiting does now work in batch mode as well and is still
enabled by default.
* The old UndoTokenWithSize CRTP hack could be removed.
UndoStack operations should be a bit faster now.
But on the other hand, there will be an overhead due to repeated
memory limit checking on every processed character.
|
|
* test case: rubout 1U[foo]
* this probably also leaked memory if it didn't crash
* a missing cast from RBTree::remove() was missing.
This cast is necessary since QRegister uses multiple inheritance.
The offset of RBEntryString might not be 0 in QRegister.
Also, since the base class is no longer virtual, a cast to the
virtual QRegister class is necessary to ensure that subclass
destructors get called.
This might have not caused problems before since RBEntry was virtual
or the compiler just happened to reorder the instance structures.
|
|
implementation classes
* whenever the implementation class was not exactly RBEntryType,
it had to have a virtual destructor since RBTree cared about
cleanup and had to delete its members.
* Since it does not allocate them, it is consistent to remove RBTree::clear().
The destructor now only checks that subclasses have cleaned up.
Implementing cleanup in the subclasses is trivial.
* Consequently, RBEntryString no longer has to be virtual.
HelpIndex and GotoTables are completely non-virtual now
which saves memory (and a bit of cleanup speed).
For QRegister, not much changes, though.
|
|
* Using a common implementation in RBTreeString::auto_complete().
This is very efficient even for very huge tables since only
an O(log(n)) lookup is required and then all entries with a matching
prefix are iterated. Worst-case complexity is still O(n), since all
entries may be legitimate completions.
If necessary, the number of matching entries could be restricted, though.
* Auto completes short and long Q-Reg names.
Short names are "case-insensitive" (since they are upper-cased).
Long specs are terminated with a closing bracket.
* Long spec completions may have problems with names containing
funny characters since they may be misinterpreted as string building
characters or contain braces. All the auto-completions suffered from
this problem already (see TODO).
* This greatly simplifies investigating the Q-Register name spaces
interactively and e.g. calling macros with long names, inserting
environment registers etc.
* Goto labels are terminated with commas since they may be part
of a computed goto.
* Help topics are matched case insensitive (just like the topic
lookup itself) and are terminated with the escape character.
This greatly simplifies navigating womanpages and looking up
topics with long names.
|
|
* the old implementation tried to avoid template programming by
making the entry comparison function virtual.
* The new RBTree implementation takes a template argument with the
implementation of RBEntry. It is now partially conventional
that the template argument must be actually derived from RBTree::RBEntry
and must define a "compare" method.
* As an advantage we now get static polymorphism (avoiding virtual
calls and allowing for more compiler optimizations) and the
the RBEntry implementation no longer has to be virtual.
* The only RB-Trees actually used are string-keyed, though.
Therefore there's a common base class RBTreeString now which
defines two synonymous "key" and "name" attributes.
* The entry base class RBEntryString is virtual again because
we do not want to propagate the RBEntryType template parameter
even further and the RBTree base class needs to destroy
entries.
This might be avoided by not defining a RBTree::clear() method,
leaving this task to the implementations.
At least QRegisters have to be virtual, though.
* RBTreeString only depends on the strcmp() and strncmp() functions
used now and only case-sensitive and case-insensitive versions
are actually required, so we instantiate these templates statically
in rbtree.cpp.
This means there are still only two instantiations of the RBTree
in the binary.
* RBTreeString defines convenient wrappers for find() and nfind()
to look up by string.
This uses the RBEntryString base class, so no allocations whatsover
are required for lookups and less space is wasted on the call stack.
* A RBEntryOwnString base class is also provided which frees the
implementations from memory managing the tree keys.
* RBTreeString can now be used to add other common functionality
like auto-completions for Q-Registers, goto labels and help topics.
* some minor optimizations
* updated TODO
|
|
* mapped to different registers beginning with "~"
* on supported platforms accessing the clipboard is as easy as
X~ or G~.
Naturally this also allows clipboards to be pasted in
string arguments/insertions (^EQ~).
* Currently, Gtk+, PDCurses and ncurses/XTerm are supported.
For XTerm clipboard support, users must set 0,256ED to enable
it since we cannot check for XTerm window ops programmatically
(at least without libX11).
* When clipboard regs exist, the clipboard can also be deemed functional.
This allows macros to fall back to xclip(1) if necessary.
* EOL handling has been moved into a new file eol.c and eol.h.
EOL translation no longer depends on GIOChannels but can be
memory-backed as well.
|
|
batch mode
* by using variadic templates, UndoStack::push() is now responsible
for allocating undo tokens. This is avoided in batch mode.
* The old UndoStack::push(UndoToken *) method has been made private
to avoid confusion around UndoStack's API.
The old UndoStack::push() no longer needs to handle !undo.enabled,
but at least asserts on it.
* C++11 support is now required, so variadic templates can be used.
This could have also been done using manual undo.enabled checks;
or using multiple versions of the template with different numbers
of template arguments.
The latter could be done if we one day have to support a non-C++11
compiler.
However since we're depending on GCC 4.4, variadic template use should
be OK.
Clang supports it since v2.9.
* Sometimes, undo token pushing passed ownership of some memory
to the undo token. The old behaviour was relied on to reclaim the
memory even in batch mode -- the undo token was always deleted.
To avoid leaks or repeated manual undo.enabled checking,
another method UndoStack::push_own() had to be
introduced that makes sure that an undo token is always created.
In batch mode (!undo.enabled), this will however create the object
on the stack which is much cheaper than using `new`.
* Having to know which kind of undo token is to be pushed (taking ownership
or not) is inconvenient. It may be better to add static methods to
the UndoToken classes that can take care of reclaiming memory.
* Benchmarking certain SciTECO scripts have shown 50% (!!!) speed increases
at the highest possible optimization level (-O3 -mtune=native -march=native).
|
|
|
|
* the problem comes from StateExpectQReg resetting the QRegMachine
too early. StateExpectQReg(QREG_OPTIONAL) states cannot call
machine.fail() in their got_register() callback.
In other words, commands with both optional or required registers
depending on runtime state cannot be modelled with StateExpectQReg.
* instead we derive from State directly - most functionality is
encapsulated in QRegSpecMachine anyway.
* might also fix crashes on some systems.
|
|
* added a new OPTIONAL behaviour for QRegSpecMachines
* allows you to implement commands that have an optional Q-Register
argument that should not be initialized if undefined.
* Using QRegSpecMachine::fail() you may still check for existence of
the register conditionally to emulate the QREG_REQUIRED behaviour.
* Using :Q for checking for register existence makes sense, because
usually you will want to check for both existence and non-emptyness
as in :Qq">. So in this common case, you no longer have to
keep in mind that the register may also be undefined.
* This finally allows us to create arrays in the Q-Register
tables without keeping a separate entry for the number of elements.
E.g. an array.0 to array.N can be iterated like this:
0Ui <:Q[array.^E\i]:; ! work with element i ! %i>
|
|
* this means that QRegSpecMachine::input() no longer has to return
a dummy QRegister in parse-only mode.
This saves an unnecessary QRegister table lookup and speeds up
parsing.
* QRegSpecMachine can now be easily extended to behave differently
when returning a Q-Register, e.g. simply returning NULL if a register
does not exist, or returning a register by prefix.
This is important for some planned commands.
* StateExpectQReg::got_register() now gets a QRegister *.
It can theoretically be NULL - still we don't have to check
for NULL in most cases since NULL is only passed in parse-only mode.
|
|
* relies on a patched version of Scinterm that allows you to
construct Scintilla objects, send messages etc. before Curses
is initialized.
The Scintilla and Scinterm submodules have been updated.
* This once and for all fixes batch mode and stdio redirections
in batch mode on all Curses platforms and operating systems.
* Fixes the ^C-does-not-interrupt bug on ncurses/UNIX.
See #4.
* On ncurses/UNIX we will still do a newterm()-initialization.
This allows us to keep stdout/stderr alone in case they are
redirected. This effectively allows redirecting SciTECO's
output into a file even in interactive mode.
ncurses/UNIX now behaves like, e.g. PDCurses/win32a and GTK+
in this regard.
* Curses environment variable handling fixed.
The environment registers are exported into the process environment
so that Curses environment variables can be set/modified by the
SciTECO profile.
* Use term.h for accessing terminfo now.
Explained set_window_title() limitations.
* fixed interruption via SIGINT. If the UI is waiting for user
input, SIGINT is effectively ignored instead of letting the
next character fail always.
* Updated sciteco(1) and sciteco(7): More options, environment variables
and signals documented. Also rewritten DESCRIPTION section
(different modes of operation).
|
|
* the registers beginning with "$" are exported into sub-process
environments. Therefore macros can now modify the environment
(variables) of commands executed via EC/EG.
A variable can be modified temporarily, e.g.:
[[$FOO] ^U[$FOO]bar$ EC...$ ][$FOO]
* SciTECO accesses the global environment registers instead of
using g_getenv(). Therefore now, tilde-expansion will always
use the current value of the "$HOME" register.
Previously, both register and environment variable could diverge.
* This effectively fully maps the process environment to a subset of
Q-Registers beginning with "$".
* This hasn't been implemented by mapping those registers to
special implementations that updates the process environment
directly, since g_setenv() is non-thread-safe on UNIX
and we're expected to have threads soon - at least in the GTK+ UI.
|
|
* expands to the value of $HOME (the env variable instead of
the register which currently makes a slight difference).
* supported for tab-completions
* supported for all file-name accepting commands.
The expansion is done centrally in StateExpectFile::done().
A new virtual method StateExpectFile::got_file() has been
introduced to pass the expanded/processed file name to
command implementations.
* sciteco(7) has been updated: There is now a separate section
on file name arguments and file name handling in SciTECO.
This information is important but has been scattered across
the document previously.
* optimized is_glob_pattern() in glob.h
|
|
appending to "$"
* these operations are unsupported and there is no benefit
in ignoring them silently. It only confused the user.
|
|
working directory
* FG stands for "Folder Go"
* FG behaves similar to a Unix shell `cd`.
Without arguments, it changes to the $HOME directory.
* The $HOME directory was previously only used by $SCITECOCONFIG on Unix.
Now it is documented on its own, since the HOME directory should also
be configurable on Windows - e.g. to adapt SciTECO to a MinGW or Cygwin
installation.
HOME is initialized just like the other environment variables.
This also means that now, the $HOME Q-Register is always defined
and can be used by platform-agnostic macros.
* FG uses a new kind of tab-completion: for directories only.
It would be annoying to complete the FG command after every
directory, so this tab-completion does not close the command
automatically. Theoretically, it would be possible to close
the command after completing a directory with no subdirectories,
but this is not supported currently.
* Filename arguments are no longer completed with " " if {} escaping
is in place as this brings no benefit. Instead no completion character
is inserted for this escape mode.
* "$" was mapped to the current directory to support an elegant way to
insert/get the current directory.
Also this allows the idiom "[$ FG...new_dir...$ ]$" for changing
the current directory temporarily.
* The Q-Register stack was extended to support restoring the string
part of special Q-Registers (that overwrite the default functionality)
when using the "[$" and "]$" commands.
* fixed minor typos (american spelling)
|
|
* activated via bit 4 of the ED flag (enabled by default)
* automatic EOL guessing on file loading and translation to LFs.
* works with files that have inconsistent EOL sequences.
* automatic translation to original EOL sequences on file saving
* works with inconsistent EOL sequences in the buffer.
This should usually not happen if the file was read in with
automatic EOL translation enabled.
* also works with the EC and EG commands
* performance is OK, depending on the file being translated.
When reading files with UNIX EOLs, the overhead is minimal
typically-sized files. For DOS EOLs the overhead is larger
but still acceptable.
* Return (line feed) is now an immediate editing command.
This centralizes EOL sequence insertion.
Later, other features like auto-indent could be added to
the editing command.
* get_eol() has been moved to main.cpp (now called
get_eol_seq()
* Warn if file ownership could not be preserved when
saving files.
* IOView has been almost completely rewritten based
on GIOChannels. The EOL translation code is also in IOView.
|
|
|
|
cleanup/refactoring
* characters rubbed out are not totally removed from the command line,
but only from the *effective* command line.
* The rubbed out command line is displayed after the command line cursor.
On Curses it is grey and underlined.
* When characters are inserted that are on the rubbed out part of the command line,
the cursor simply moves forward.
NOTE: There's currently no immediate editing command for reinserting the
next character/word from the rubbed out command line.
* Characters resulting in errors are no longer simply discarded but rubbed out,
so they will stay in the rubbed out part of the command line, reminding you
which character caused the error.
* Improved Cmdline formatting on Curses UI:
* Asterisk is printed bold
* Control characters are printed in REVERSE style, similar to what
Scinterm does. The controll character formatting has thus been moved
from macro_echo() in cmdline.cpp to the UI implementations.
* Updated the GTK+ UI (UNTESTED): I did only, the most important API
adaptions. The command line still does not use any colors.
* Refactored entire command line handling:
* The command line is now a class (Cmdline), and most functions
in cmdline.cpp have been converted to methods.
* Esp. process_edit_cmd() (now Cmdline::process_edit_cmd()) has been
simplified. There is no longer the possibility of a buffer overflow
because of static insertion buffer sizes
* Cleaned up usage of the cmdline_pos variable (now Cmdline::pc) which
is really a program counter that used a different origin as macro_pc
which was really confusing.
* The new Cmdline class is theoretically 8-bit clean. However all of this
will change again when we introduce Scintilla views for the command line.
* Added 8-bit clean (null-byte aware) versions of QRegisterData::set_string()
and QRegisterData::append_string()
|
|
* acts as a safe-guard against uninterrupted infinite loops
or other operations that are costly to undo in interactive mode.
If we're out of memory, it is usually too late to react properly.
This implementation tries to avoid OOMs due to SciTECO behaviour.
We cannot fully exclude the chance of an OOM error.
* The undo stack size is only approximated using the
UndoToken::get_size() method.
Other ways to measure the exact amount of allocated heap
(including size fields in every heap object or using sbrk(0) and
similar) are either costly in terms of memory or platform-specific.
This implementation does not need any additional memory per heap
object or undo token but exploits the fact that undo tokens
are virtual already. The size of an undo token is determined
at compile time.
* Default memory limit of 500mb should be OK for most people.
* The current limit can be queried with "2EJ" and set with <x>,2EJ.
This also works interactively (a bit tricky!)
* Limiting can be disabled. In this case, undo token processing
is a bit faster.
* closes #3
|
|
* also did some whitespace cleanup in SciTECO now that tabs are
displayed properly
|
|
|
|
|
|
* EW can save Q-Registers now
* the new E% may be used to save a q-register without making it
the current document
|
|
it became apparent, that something like this is very useful,
when constructing the contents of a q-register without
editing it.
I have decided against introducing another modifier for toggling
string building. Most commands have string building enabled and it
doesn't hurt. For the few exceptions, an alternative variant can
be introduced.
|
|
this is more consistent with SciTECO's idea of abstract registers
and allows the currend buffer to be saved on the Q-Register stack.
This allows the idiom: [* ! ...change current buffer... ! ]*
|
|
This is only a problem if the macro created the local Q-Register table
(i.e. not when called with ":M") but resulted in segfaults.
Since we do not want to save in a Q-Reg whether it is local
(and that wouldn't suffice anyway), we do it in the Q-Register table
cleanup. The corresponding QRegisterTable::clear() must be called
explicitly, since the RBTree::clear() called on destruction does not
and cannot throw errors.
If QRegisterTable::clear() has been called successfully, the default object
destructor will not do much. If it has thrown an error, the destructor
will clean up the remaining Q-Registers.
|
|
* it must be initialized after the UI (Interface::main), so I added
a View::initialize() function
* the old initialize() method was renamed to setup()
* use a global instance of QRegister::view so it is guaranteed to
be destroyed only after any QRegisters that could still need it
* Document API adapted to work with ViewCurrent references
|
|
* allowed me to remove some obscure global functions and methods like
QRegister::update_string().
* Document updating is concentrated in qregisters.cpp now
* also fixes some bugs introduced earlier, like undo tokens being
generated for non-undo registers (resulting in segfaults on rubout)
|
|
The user interface provides a Scintilla view abstraction and
every buffer is based on a view. All Q-Register strings use
a single dedicated view to save memory and initialization time
when using many string registers.
* this means we can finally implement a working lexer configuration
and it only has to be done once when the buffer is first added
to the ring. It is unnecessary to magically restore the lexer
styles upon rubout of EB (very hard to implement anyway). It
is also not necessary to rerun the lexer configuration macro
upon rubout which would be hard to reconcile with SciTECO's
basic design since every side-effect should be attached to a
character.
* this means that opening buffers is slightly slower now
because of the view initialization
* on the other hand, macros with many string q-reg operations
are faster now, since the document must no longer be changed
on the buffer's view and restored later on.
* also now we can make a difference between editing a document
in a view and changing the current view, which reduces UI calls
* the Document class has been retained as an abstraction about
Scintilla documents, used by QRegister Strings.
It had to be made virtual, so the view on which the document
is created can be specified by a virtual function.
There is no additional space overhead for Documents.
|
|
normally, since SciTECO is not a library, this is not strictly
necessary since every library should use proper name prefixes
or namespaces for all global declarations to avoid name clashes.
However
* you cannot always rely on that
* Scintilla does violate the practice of using prefixes or namespaces.
The public APIs are OK, but it does define global functions/methods,
e.g. for "Document" that clashed with SciTECO's "TECODocument" class at
link-time.
Scintilla can put its definitions in a namespace, but this feature
cannot be easily enabled without patching Scintilla.
* a "SciTECO" namespace will be necessary if "SciTECO" is ever to be
turned into a library. Even if this library will have only a C-linkage
API, it must ensure it doesn't clutter the global namespace.
So the old "TECODocument" class was renamed back to "Document"
(SciTECO::Document).
|
|
the expression stack
now it's more like standard TECO's ^U command
|
|
* referencing temporaries is unreliable/buggy in GNU C++, at least since v4.7
* in higher optimization levels it resulted in massive memory corruptions
* this is responsible for the build issues (PPA build issues)
* instead, always declare a buffer on the stack which guarantees that the
variable lives long enough
* the g_strdup(CHR2STR(x)) idiom has been replaced with String::chrdup(x)
|
|
|
|
* results in better error messages, e.g. when opening files
* the case that a file to be opened (EB) exists but is not readably is handled for the first time
|
|
* specifications resulted in runtime errors (unexpected exception) when bad_alloc ocurred
* specs should be used scarcely: only when the errors that may be thrown are all known
and for documentary purposes
|
|
* when an error is thrown, stack frames are collected on clean up, up to
the toplevel macro
* the toplevel macro decides how to display the error
* now errors in interactive and batch mode are displayed differently
* in batch mode, a backtrace is displayed as a sequence of messages
* Execute::file() forwards errors correctly
* the correct error in the file is displayed in interactive mode
* necessary to build the stack trace
|
|
* the warning itself makes sense but in the cases reportet they
were irrelevant
|
|
|
|
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
|
|
tab-completions
* StateExpectFile adds no functionality (currently), but is useful for checking state types
|
|
* useful for using macro libraries
|
|
* 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
|
|
* distinction no longer useful since string part of register
is now never pre-initialized
|
|
* 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.
|
|
|
|
* enabled for all modifying Q-Reg commands
|
|
syntax is as follows: ["."]("#" CHR1 CHR2 | "{" STRING_BUILDING "}" | CHR1)
* the short one/two char names are turned upper case, while no
case folding is performed on verbose names
|
|
occurs when rubbing out a switch from q-reg string or to q-reg string
|
|
allows commandline editing scripted by macros
|
