Previous implementation doesn't care about characters that are not
covered by attributes run length array elements - so the attributes of
the next run array element was taken instead of default one on the
HistoryBuffer::GetTerminalLine() call. Note that this was the case of
"holes" in run array but not the characters after the end of run array
elements.
* Lot of fixes to add support of BCE (background color erase). Shell
is switched to emulate xterm-256colors terminal, that is modern and
declare the colors capabilities of our Terminal more precisely;
* Move current character attributes fAttr from TermView to
BasicTerminalBuffer. This reduces count of function parameters on call
various InsertXXX routines;
* In alternative screen buffer mode the whole cells "matrix" of the
screen buffers is taken into account during drawing background of the
cells in the view. In normal mode the "attributes" field of the
TerminalLine is used to detect color of the area after the last
character - there should be no changes with previous behaviour;
* Fix attributes on kSpaceChar-padding short lines. Current _line_
attributes should be used instead of current _global_ attributes;
* Fixed pads and gaps attributes, more accurate handling of ESC[K and K°;
* _Invalidate strings just erased. Fix EraseChars DCH processing;
* Fixes for ESC[J erase lines control sequences;
* Added handling SGR 90-97, 100-107 codes;
* Clean the newly allocated TerminalLine lines;
* More precise cursor background [off-]color estimation at ends of
lines. The current line attributes should be used instead of
hard-coding it to fTextBackground;
* Fixed background color erase in normal screen buffer modes. Wrong
line indexes calculation messed the drawing results or just returned
0 [default] line attributes;
* Some more BCE support: TerminalLine::Clear() now honors current
character attributes;
* Fixes#6143, #6510 and #6424.
Two helper functions introduced: "make debug snapshots" triggered by
Ctrl-Cmd-S shortcut and "capture data flow" triggered by Ctrl-Cmd-C
shortcut. The first one makes debug dump of current data both in visual
and in text buffer, including history lines if they available. The second
one mirrors all characters and control sequences that are flowing
through the Parser. Both dump and capture files are saved under /var/log
folder. That functionality available only if the USE_DEBUG_SNAPSHOTS switch is
defined.
This reverts hrev38316 in parts of foreground and background colors
handling. Mentioned revision has hardcoded colors to system ANSI color
entries 7 and 0 correspondently. It is very uncomfortable for the
console programs like Midnight Commander because they become
colors that were defined in the current Terminal scheme and doesn't
correspond to real ANSI color names. For example "ANSI black" can be
black, white, blue, gray or something else in dependency of the current
color scheme configuration. The same side-effect was obviously observed
for "ANSI white" and produced funny color combinations in software.
* Global kTermColorTable replaced with it's private copy for every
TermView instance. This allows to modify colors table for every view
separately;
* Set of ANSI normal/bright color entries added into preferences;
* Default color table generated dynamicaly using ANSI colors
preferences. 6x6x6 color cubes and grayscale ramp are generated in
xterm colors model;
* Improoved support of Operating System Command control sequences;
* Support for X11 rgb.txt compatible color names. Corresponding entries
are stored in hashed form in application resource and loaded only on
demand.
* PREF_SHELL is not used anymore, corresponding entry in /etc/passwd
should be used instead;
* PREF_GUI_LANGUAGE is not actual anymore - we have system-wide
UI localization support;
* Fix typo in the name of MSG_SET_TERMNAL_TITLE constant.
Terminal is GUI application and there are no need to emulate bold
in some kind of color adjustments, especially in case the proposed
emulation look is too far away from results observed on similar software
like xterm, Konsole and XFCE Terminal.
New version of termcap.src was generated using the "master" terminfo.src
database taken from the ncurses 5.9 package. This database is looking
like only one up-to-date I have found. This update is pre-requisite for
lot of Terminal BCE refactoring changes that follow soon;
File was generated using "tic -CrtT" command and was fixed for unpaired
:fs= and :sp= entries to satisfy ncurses' sanity check.
* HttpAuthentication.cpp wasn't in ssl grist
resulting in unmet dependencies. #9523
* md5.c only being used when ssl wasn't available
was kind of spaghetti logic. These changes make
this more obvious.
Added the optional package for system sounds that
were collected for GCI 2012. Also added the demo
packges to the "contents" at the top of OptionalPackages.
The standard states that F_GETLK should check whether given lock would be
blocked by another one and return description of the conflicting one (or
set l_type to F_UNLCK if there is no collision).
Current implementation of F_GETLK performs completely different actions, it
"Retrieves the first lock that has been set by the current team". Moreover,
if there are no locks (advisory_locking == NULL) an error is returned
instead of l_type set to F_UNLCK.
Recently enabled variable range propagation enables GCC to among others analyze
whether array subscript is in a valid range. While being quite useful this
particular feature also happens to produce false positives. This is merely
a workaround to make compiler happy.
The actual reason why compiler is reporting false positive is that array is
indexed with a signed integer and it is not clear for compiler what value it may
have due to it being a member of the class and and external procedure being
invoked between its initialization and usage.
* This can be done from the command line using the new "-open" option.
* When the application is invoked via Tracker, ie. by double clicking
on an image, then it's being used implicitly.
* For LVDS output, the sync parameters aren't really used which is why the mode
set from the BIOS might not be valid if the PLL hardware would actually be
used.
* Therefore, we sanitize this mode to make sure it's within allowed parameters
so that intel_set_mode() will accept it.
* This should fix at least #8132, and #8796.
* Sorry for looking into this so late! I obviously messed up testing this
before, as my EeePC 900 was actually affected by this, too.
* Replace {Set|Remove}MasterKey() by generic {Set|Remove}UnlockKey()
that works on a keyring.
* Implement {Set|Remove}MasterUnlockKey() on top of that.
* Rename the commands and constants accrodingly.
* Implement setting and removing keyring unlock keys.
* Rename fKeyMessage to fUnlockKey and the KeyMessage() getter to
UnlockKey().
* Keep track of whether the keyring has an unlock key set.
* Store and restore that info separately.
* En- and decryption will depend on unlock key presence later.
* Add functions to set and remove an unlock key and query for it.
* Each keyring is now stored in a proper message which allows it to
contain additional meta data along side the flat data.
* Adding all keyring messages under a common field also allows to add
meta data to the keystore, as the keyrings don't use up random field
names anymore.
* Treat the master keyring as any other keyring and just add it to the
list. This allows to write/read the keystore database without special
casing the master keyring.
As there aren't any more generic meta data containers inside BKey,
there's no real way to distinguish different instances with the same
identifiers. This may be added later, for example the same index system
as used in BMessage could apply.
Before, each permanently granted access flag would overwrite the
previously granted flag, causing the dialog to come up whenever the
operation was changed.
