There are more independent crashes occuring in this scenario and this
fix targets only one of them, in particular the one in which different
window's outliner sends outliner command to a wrong window. The fix
itself gives an option for lfun to know which window it belongs to.
https://www.mail-archive.com/lyx-devel@lists.lyx.org/msg203619.html
We now report whether the attempt to recode the macros to glyphs
succeeded. If yes, we set "literate" to false, if not to true.
Also, do not attempt to recode for non-latexifying fields.
Fixes: #9563
This requires moving the bool that tracks this somewhere that it
is visible from BufferView. It seemed to make sense to put it as
a static member of InsetCitation.
Things like
pdfpagemode=UseOutlines%None,UseOutlines,UseThumbs,FullScreen
was not imported correctly (the comment was not stripped)
Fixes the rest of #5737
Add a `cancel' boolean to macroModeClose() that just removes all trace
of what has been entered instead of finalizing it.
When entering a macro in mathed, let LFUN_ESCAPE invoke
macroModeClose(true). The new semantics of LFUN_ESCAPE in mathed is
thus to abort the input of a macro name.
Fixes bug #9251.
Use proper value for above/below skip (12pt) instead of the hardcoded
12 pixels.
Add forgotten skip before display math preview.
Make "too small preview" size dpi-dependent.
Make gap in front of equation (what is it good for, BTW?) dpi-dependent.
As of v. 1.7.7, chktex has four exit values. Only consider the program
failed with EXIT_FAILURE (1). This is backwards compatible to chktex
up to v. 1.7.5 and later patched versions included in TeXLive, where
there was the distinction EXIT_FAILURE (program failed) and EXIT_SUCCESS
(program successfully run, with or without something to report).
Note that ChkTeX v. 1.7.5 and 1.7.6 vanilla (as included in MikTeX) also
returned EXITE_FAILURE if ChkTeX found something to report.
We do not, and never did, support this case.
Fixes: #9989 (after ChkTeX 1.7.7. is released).
In cursorY, it is dangerous to access par_petrics_[0], since one does
not know whether metrics have been computed for this paragraph (which
may be off-screen).
It is safer to use parMetrics(0), that will compute the paragraph
metrics as needed.
Fixes bug #8120.
I am rather irritated we didn't do this already, since synchronous runs
with BUFFER_VIEW or BUFFER_UPDATE leads to all sorts of problems,
including crashes.
Fixes the crash in #8338 (but not the bug itself).
Since CheckTeX itself processes the tex file, a synchronous run with
a TeX process can lead to all sorts of conflicts, including crashes.
Fixes: #7434.
The issue here was that the element was only removed from the queue
after the func request was processed, but within that process, other
function could access the queue, so the queue could even be empty
when this function finally wanted to remove the item.
Fixes: #10406.
When pasting "\big" without any following delimiter, avoid
processing the same token again and again. For unknown reasons,
the delim docstring turns out to always be not empty: even when
it simply contains a '0' (no delimiter follows), its length is 1.
Fixes bug #11027.
Several problems:
* The regex failed at names such as 1_text_2_text.tex
(returned "2_text.tex)
* The regex failed at names such as 12_text.tex
(returned "2_text.tex)
* Masters with digits in the name (2018_text.tex) were
tracked as their own children
This is needed for cases where the temp file has to be manually removed
at some point (e.g., if temp files are used as conversion target, and
the initial file only serves as a placeholder), since QTemporaryFile
objects cannot be manually removed at least on Windows (they are always
kept open internally even after close()). See
http://lists.qt-project.org/pipermail/interest/2013-August/008352.html
In order to avoid race conditions due to duplicate names (the issue why
the old method was removed), we record all used temp file names.
Fixes: #9139
This is a follow-up to 02028c0b12.
Fix two bugs introduced when moving Cursor::sanitize() to CursorData:
* the setBuffer that is supposed to set buffer from bufferview became
a no-op.
* new_word_ is now part of CursorData, but it is sanitized only in
Cursor::sanitize().
Additionally, make all protected CursorData data members private to enforce
encapsulation.
Before accessing things like new_word_.lastpos(), it is better to make
sure that new_word_ points to something that exists. Therefore, the
call to fixIfBroken() is moved in first position.
Other changes: inTexted() is replaced by a more general test that
current inset has not changed; test idx() in addition to pit().
Use the LaTeX internal character representation (LICR) macros
provided by lgrenc.def (since version 0.8 from 2013-05-13)
in lib/unicodesymbols. This fixes the PDF bookmarks (except for the
legacy input encoding iso-8859-7) and solves the problem of a missing
"v" character in Libertine LGR fonts (see lyx-users from 2018-01-29).
The ctest unicodesymbols/008-greek-and-coptic_iso8859-7_pdf2" now fails
(due to #9681). This is not a regression, as it is already
"unreliable" (wrong output, Latin character instead of Greek).
Drop compatibility definition of \~ as perispomeni accent
(that was required with lgrenc.def < 0.8).
As of enchant 2.x, it is required to create a Broker instance instead
of relying on a static one provided by the library.
Add autoconf and cmake (courtesy of Kornel) tests that check whether
one can indeed instantiate a Broker object, and act on the result in a
new broker() helper function.
Fixes bug #10986.
Actually, the possible exception path is from replaceEnvironmentPath.
It would be better if coverity was able to asses that our hardcoded
regexp:s are correct.
Now people know for sure that I sometimes forget to test that code
actually works before committing. It was so straightforward, what
could possibly go wrong?
This is currently only relevant fo InsetListings, which falls back to
a fixed-width encoding under specific conditions. It is now possible
to query the inset about that and report the correct encoding in
DocIterator::getEncoding.
Addresses the second part of #10995
The function does not use for now any information from the BufferView
(only lyxrc), but this should eventually change if we want to honor
multi monitor setups properly.
On platforms where char_type is typedef'd to an integral type
rather than to wchar_t, when using the insertion operator <<
a single char_type is output as the number corresponding to the
code point of the character rather than as the character itself.
In this case, one has to use put().
Current code would change the color of static endlabel when the
_start_ of line is selected. This is only used with hollywood and
broadway layouts, though.
Change the box endlabel painting code to use Color_selectiontext as needed.
Follow-up to bug #10972.
This is preliminary work, this code still feels too complicated for
its own good.
Let Row::isMarginSelected return false when Row::selection() is false
(the other changes are indentation).
This allows to remove the test for selection() in
setSelectionAndMargins, so that begin/end_margin_sel are always set
correctly.
Add clearSelectionAndMargins() instead of calling directly setSelection
(which is now private) with arguments (-1, -1).
Fixes bug #10972.
A brand new event compressor based on Kuba Ober's cleverly simple
solution: <https://stackoverflow.com/a/21006207>.
Fix#9362, #9461, #9933: Lyx suddenly gets keyboard keys wrong, and
deadlocks
Fix#9790: LyX should perform key event compression (for improving the
remote X connections one would also need to implement
Qt::WA_KeyCompression)
Fix#10516: slowness on repeated arrow keys with IBus and Qt5
Patch pulled from
https://github.com/gadmm/lyx-unstable/commit/bf5a1efb0db5bfc2b
Signed-off-by: Juergen Spitzmueller <spitz@lyx.org>
This fixes a regression in e64ea357, where a test has been (badly)
tightened to avoid that two consecutive rows may be redrawn to get rid
of caret ghosts. The test prohibited empty rows from being redrawn.
Moreover, improve the test of cursor boundary to avoid the case where
cursor position is already 0.
Fixes bug #10952.
A basic class like Length should not depend on something from
frontend.
This change allows to remove several dummy implementations of
theFontMetrics().
At some time it seemed like a good idea in breakRow() to return early
when the row was bound to be empty. It turns out that this creates two
symptoms:
* empty paragraphs will not have an end of paragraph marker
* since row width is not correctly computed in this case, caret ghosts
can appear in master.
This commit removes the oprimization and replace the do {} while()
construct to a straightforward while() {}.
Related to bug #10952.
It is not good for a support function to depend of anything outside of
it, especially Application.
Here the boolean that indicates that export should be canceled is put
in Systemcall. This allows to remove all the dummy theApp() function
that have been added here and there for linking needs.
Ligating of -- to en dashes occure also in teletype fonts.
With some 8-bit fonts, em and en dashes are not copied
exported from the PDF (but this is not limited to dashes in teletype).
With LatinModern, PDF export works fine
MWE:
\documentclass[]{article}
%\usepackage{lmodern}
\usepackage[T1]{fontenc}
\begin{document}
Hallo \texttt{Welt --Welt ---Welt}
Hallo Welt --Welt ---Welt
This gives a more consistent indication of what belongs to an environment
(and semantically, such paragraphs are synonymous to nested standard
paragraphs)
Addresses part of #9261.
VCS InsetInfos were broken by the switch at 2e934fc5f8 to using
updateBuffer to handle them. But we do not really want to go through
that routine in a clone, not for those insets, whose contents we
want to be the same as in the original Buffer.
Also fixes some issues noted in discussion of this bug: Failure to
update after context menu switch; failure to re-calculate shortcuts,
which can change.
This is complementary to AutoNests: Styles can determine which other
style should auto-nest them. This is particularly useful for modules
that add new styles which should be auto-nested in a given context.
Now layouts can specify other layouts that should be nested in and
after the current one (if the layout is switched from the current one
and if it follows a paragraph in the current one).
This is particularly useful for things such as the beamer frames, where
particular layouts are practically always nested.
These classes come with their own working implementation of table
footnotes, so we are not supposed to use the tablefootnote or
footnote package.
Also add missing outliner name.
This gets rid of the hardcoded latin1 encoding for verbatim. Instead,
verbatim now inherits the encoding from the context, which is what is
actually wanted here.
Fixes: #9012, #9258
While there is code in LyX that handles this, tex2lyx should respect
the format that LyX produces and the Flex: prefix is gone since
cfeddb92 in 2014.
In particular, this prefix is known to cause assertion when undoing a
document setting change.
This had the effect that for file dialogs supposed to use two buttons,
only the second one was displayed, while file dialogs supposed to use
one button did not have a button at all.
Candidate for 2.3.x and 2.2.x.
This spacing was wrongly removed at 361bd53b as part of the
introduction of proper spacing btween elements, but it only introduces
some breathing space around the delimiters. This will not be necessary
anymore when/if we use proper delimiters from the cmex10 font.
By default, inactive math corners are invisible. In practice they are
annoying because they are visible when selecting text, and they can
also overwrite some parts of the equation.
The code in Inset::drawMarkers2, which is only used for maths, is
moved to InsetMathHull. Moreover, the inactive corners are not drawn
when they have the same color as the math background. A better way to
achieve this would be to set the color to transparent, but we do not
support this yet.
Qt on macOS does not respect the Qt::WA_OpaquePaintEvent attribute and
clears the widget backing store at each update. Therefore, we use our
own backing store in this case.
This restores a simplified version of the code that was removed at 24c29908.
Now the helper class contains logic that checks whether buffer are
known before closing them. This avoids potential crashes.
Use it in different places to siplify code. It is not clear at this
point whether it should be used everywhere.
Followup to bug #10847.
When a buffer is in an embedded work area (adv. find&replace), it is
not found by BufferList:::exists(), and therefore the undo group
created in GuiApplication::dispatch and in the handling of
LFUN_COMMAND_SEQUENCE will not be closed.. Crashes can ensue, as
described in Ubuntu bug:
https://bugs.launchpad.net/bugs/1737429
The solution is to introduce BufferList::isInternal and act on it.
Fixes bug #10847.
This is a follow-up to 758de957.
- unify the metrics and drawing of \sqrt and \root using helper
functions mathed_root_metrics and mathed_draw_root.
- compute the vertical spacing above the nucleus of the root following
rule 11 of the TeXbook. In particular, it is different in inline and
display style.
- draw the root glyph without hard-coded pixel values. Make the line
width depend on the zoom.
more work is needed to implement properly rule 11:
- Ideally, we should use sqrt glyphs from the math fonts. Note that
then we would get rule thickness from there.
- The positioning of the root MathData is arbitrary. It should
follow the definition of \root...\of... in The Texbook in
Apprendix B page 360.
Fixes bug #10814.
It may happen that the buffers are visited in order buffer1,
buffer2, buffer1. In this case, we want to have only one undo group
in buffer1. The solution is to replace buffer_ with a set<Buffer*>.
A use case among others is InsetLabel::updateReferences.
The bug is the following: when selecting several paragraphs quickly
enough, some rows do not get selected.This is a consequence of the
removal of row crc, which lead to not taking into account the
selection status of the row in the decision to repaint.
The solution chosen here is to add a Row::change() helper function to
modify row members. This will set the Row changed status whenever the
value of the member changes.
The correct behavior is to go to position 0 going up from first row,
and to end of row when going down on last row.
The targetx value of the cursor is not updated, which makes cursor
movement more natural.
Fixes bug #10701.
When the caret is at end of row, if may happen that it is drawn after
the end of the row. In this case caret blinking will not work
properly. This patch extends the row background on the left and right
by Inset::TEXT_TO_INSET_OFFSET. This is only a hack that will not work
if the caret has a ridiculous width like 6.
Additionally, introduce some (disabled) debug code that numbers the
rows on screen by painting order.
Finally, make the code that detects whether the caret was in a given
row more precise (take boundary into account).
Fixes (mostly, see above) bug #10797.
This computation did not make sense anymore since we began to put the
contents in the Row object. The fact that it worked was a coincidence.
Instead, we set rows as changed() on creation and reset that once they
have been drawn. This will allow in the future for a finer definition
of what to redraw or not.
Also update the PAINTING_ANALYSIS document
It is wrong to compute this at paint time. In general, painting a row
should not require any access to a paragraph object, but we are far
from there now.
Along the lines suggested by JMarc, we now collect the list of bibfiles
in use in the updateBuffer routines. This actually does simplify the code
quite a bit. See the discussion there for reasons to go this way.
No intended change of behavior. The ones that use DocIterators and
ParIterators cannot yet be changed, but they could be if we provided
the right sorts of methods, perhaps.
- the column width must be as wide as the column header text
- also center the radiobuttons in the table
- also use alternating colors for the table rows
Basically, everything that does not depend on a BufferView should move
there. Some methods that do not seem to need a BufferView, like
selHandle or IdxFirst or push actually depend on it and could not be
moved.
This allows to simplify a few uses of recordUndo helpers.
- Move some methods to DocIterator: nextMath, prevMath, getPossibleLabel,
getEncoding;
- Move some methods to CursorData: setCursor, setCursorSelectionTo,
(setCursorTo|normal|reset)Anchor, (set|clear)Selection,
sel(|ection)(Begin|End), selectionAsString, info, currentState,
(mark|clear|check)NewWordPosition, fixIfBroken, sanitize, all undo
related methods, reset, isInside, leaveInset, current mode;
- kill some unused methods: macromode, replaceWord, setScreenPos, touch,
markInsert, markErase;
- Move code around to group things, and add a few comments (a lot remains to be done).
This changes lead to some related changes in other classes: removal,
change of parameter.
No intended change.
The goal of this commit is to ensure that a processUpdateFlags call
that requires no redraw will not override a previous one that did
require a redraw.
To this end, the semantics of the flag argument is now different: its
value is now OR'ed with a private update_flags_ variable. This
variable is only reset after the buffer view has actually been
redrawn.
A new Update::ForceRedraw flag has been added. It requires a full
redraw but no metrics computation. It is not used in the main code
(yet), but avoids to compute metrics repeatedly in consecutive
processUpdateFlags calls.
The process is now as follows:
- if flags is just None, return immediately, there is nothing to do.
- the Force flag is honored (full metrics computation) and replaced
with ForceDraw.
- the FitCursor flag is honored and removed from the flags.
- the SinglePar update is added if ForceDraw is not in flags and only
the current par has been modified.
The remaining flags are only then added to the BufferView update
flags, and the update strategy is computed for the next paint event.
Finally the dubious call to updateMacros in updateMetrics has been
removed for performance reasons.
After 6642152e, user macros were no longer wrapped in \ensuremath. In 2.2 and
before, InsetMathMacro behaved as follow:
* Textmode global symbols are wrapped in \text when in math.
* Other global symbols, and user macros, are wrapped in \ensuremath when in
text.
* Undefined macros (ERT) are wrapped neither in \text nor in \ensuremath.
This is also consistent with the documentation of MathEnsurer in
mathed/MathStream.h.
This patch defines InsetMathMacro::currentMode() accordingly (respectively
TEXT_MODE, MATH_MODE and UNDECIDED_MODE) and uses it to determine the output.
After this patch, there is a mismatch between screen and pdf output for user
macros in \text. This is not a regression wrt 2.2 and is because linearization
does not satisfy currentMode() currently.
- If a display math not starting a new paragraph is deleted, the
current \lyxdeleted macro (if any) must be closed and a new one
started, otherwise the display math will be shifted up.
- Use \linewidth instead of \columnwidth because the former will adapt
to the reduced horizontal width in list environments, avoiding shifting
to the right the diplay math.
Spaces are, amazingly, allowed at the end of bibliography keys. So we
introduce a new parameter allowing getVectorFromString not to trim
whitespace, and then use it.
Instantiating a single QSettings and using it for each ui element
can significantly shorten the time required to save the various
states at exit. The speed up can be better appreciated on *nix,
where the settings are saved on disk, rather than on Windows where
they are held in memory (in the registry).
Scanning is rather slow, so this improves performance in specific
situations (multiple inclusion of larger files in master/child or
chapterbib context)
The goal of this commit is to ensure that a processUpdateFlags call
that requires no redraw will not override a previous one that did
require a redraw.
To this end, the semantics of the flag argument is now different: its
value is now OR'ed with a private update_flags_ variable. This
variable is only reset after the buffer view has actually been
redrawn.
A new Update::ForceRedraw flag has been added. It requires a full
redraw but no metrics computation. It is not used in the main code
(yet), but avoids to compute metrics repeatedly in consecutive
processUpdateFlags calls.
Finally the dubious call to updateMacros in updateMetrics has been
removed for performance reasons.
The current spelling is not strictly wrong, but flagged as unusual or
historical by some authorities. It is also found fault with many
spell checkers. Thus we decided to move to the more standard "-ible"
form once and for all.
See #10678 for discussion
This part covers the most tricky part: the internal naming.
Translations and layouts will follow.
This will all also all be backported to 2.3.x, for the sake of backwards
compatibility (cherry-picking).
The current spelling is not strictly wrong, but flagged as unusual or
historical by some authorities. It is also found fault with many
spell checkers. Thus we decided to move to the more standard "-ible"
form once and for all.
See #10678 for discussion
This part only covers the usage in comments and the like. More to follow.
This will all also all be backported to 2.3.x, for the sake of backwards
compatibility (cherry-picking).
in updateBuffer().
This is in response to a reported crash. See
https://www.mail-archive.com/lyx-devel@lists.lyx.org/msg202217.html
Commit af381a2f addressed the crash, and is worth doing anyway. But
this also makes sense.
Also adds a flag to tell us whether we need to recalculate the
relevant information. In some cases, there is no need to do so
more than once. (Note that, with the existing code, we *never*
recalculate, which would have given rise to bugs.)
At least with gcc 6.4, if the first parameter passed to
regex_match() is afterward changed, the second one gets
corrupted. This is avoided by using a temporary string.
* Describe the viewport/bb situation more accurate
* Use generic term "coordinates", since bb and viewport are flavor-specific
* Add some tooltips
* Increase the width of the options widget.
The previous redirection of the senseless option "bb" to "viewport" in
PDFLaTeX output has been removed in a recent graphics package update.
This breaks documents, since clipped graphics silently stop displaying.
This change restores the previous output by using "viewport" instaed of
"bb" for non-PS/DVI output, while leaving PS/DVI untouched (where "bb"
and "viewport" behave differently.
Fixes: #7910
Fix several unreleated cases where the `selecting' code path did not
reset cursor font correctly. This leads to a caret that does not have
the correct size, and can cause caret droppings with the new display
scheme.
In some insets such as Arguments, a local language switch has to be
used. However, if the language inside the inset was set to be equal
to the outer language, the code decided not to switch language. But
then got confused and tried to close a switch that was never opened.
This patch forces the switch even if the outer language is the same.
This replaces a showCaret/hideCaret pair and avoids an update.
Also remove an update() call in resizeBufferView: is is not necessary
since we are already in a pintEvent handler.
For some reason the existing code only considered the bottom row that
contained the cursor. There is no need for that, and actually it
caused painting problems in nested insets.
Tweak the logic of repaint_caret_row_ a bit: there is no need for
repainting when there is currently no caret on screen.
If the height of the first row of the document has become smaller (for
example when it contains an image which size has shrunk), that the
anchor offset has to be adapted so that there is not a large blank
space at the top of the document.
In some cases, it is possible that the BufferPositionMap has
dangling pointers. We thus check whether the Buffer is loaded
before accessing it.
Fixes: #10766
(cherry picked from commit 4f50cbcfe4)
Replace the tricky code in LFUN_SCREEN_FONT_UPDATE and replace it with
proper use of DispatchResult flags.
LFUN_BUFFER_ZOOM* does not need to call LFUN_SCREEN_FONT_UPDATE, since
it already does everything that is required.
This avoids endless resize issues on HiDPI systems (e.g. Retina Mac).
Rename pixel_ratio_ to last_pixel_ratio_ to emphasize that this is a
cached value.
Inline needResize method to make the logic clearer in paintEvent.
Remove the old schedule_redraw_ mechanism that was only useful because
of our synchronous drawing code. Now that actual painting is
scheduled instead of forced, it becomes pointless.
Rename WorkArea::redraw(bool) to scheduleRedraw(bool), to show that
the drawing is not done right away.
In GuiView::updateInset, call scheduleRedraw(true), so that metrics
are correctly computed (this was the whole point of the exercise).
This change solves dialog stacking problems on newer Mac OS X in combination with the OS provided file open dialog.
(cherry picked from commit 97dc585138)
Now painting the workarea is done at paint events as should be.
Explicit painting after updating metrics has been replaced by a much
lighter procedure (updatePosCache) to update the insets positions cache.
Expected benefits:
- better performance
- proper use of subpixel aliasing
The LyXRC variable use_qimage is not needed anymore and is therefore removed.
This patch makes sure that, every time a ParagraphMetrics has its
position set, the inset positions for the insets held by this
paragraph are remembered too.
This is complementary to BufferView::updatePosCache, but I do not have
hard evidence that this is required other than to increase robustness.
It may help in some cases when scrolling the document (scrollbar,
cursor up/down, page up/down).
This unbreaks input methods by splitting the part of the code that
does the actual drawing to a separate paintPreeditText() method which
is called from paintEvent().
The proper solution would have been to introduce the preedit string in
the Row object, like is done for completion, but this is too complex
to do at this point.
The only change in behavior is that now the commit string is inserted
in one fell swoop, intead of emulating a number of key events.
Rename cursor to caret to in order to avoid ambiguity. The caret is
now the blinking thing only.
Remove unused header contents, and some not so useful methods.
No intended change of behavior.
The trick is to remember in BufferView what has been done at the
previous draw, so that the row that contained the caret can be
repainted if needed.
To this end, add an argument paint_caret to BufferView, although
painting the caret is not the job of the BufferView (at this point).
BufferView::needRepaint will act as an interface with
TextMetrics::drawParagraph to know whether the painting of a given
row should be forced.
Currently everything is done at the top row level, so that, if the
caret is in a large table, the whole table will have to be repainted.
It is not clear yet that this is necessary.
Historically, because of two-stage drawing, LyX has been painting on a
Pixmap, and this pixmap is copied to screen at paint event time.
Now that we have three-stage drawing, it is possible to delay the
painting to actual paint event and avoid the intermediate Pixmap.
Known bug: the cursor is never erased.
