The setting of insets positions was done twice in updateMetrics.
When one of the paragraph is a huge branch, this can be very expensive.
This leads to a 17% improvement on updateMetrics time on a scrolling test.
Part of bug #12297
The idea of single par update is to try to re-break only the paragraph
containing the cursor (if this paragraph contains insets etc.,
re-breaking will recursively descend).
The existing single paragraph update mechanism was tailored to work
only at top level. Indeed changing a paragraph nested into an inset may
lead to larger changes.
This commit tries a rather naive approach that seems to work well: we
need a full redraw if either
1/ the height has changed
or
2/ the width has changed and it was equal to the text metrics width;
the goal is to catch the case of a one-row inset that grows with
its contents, but optimize the case of typing in a short paragraph
part of a larger inset.
NOTE: if only the height has changed, then it should be
possible to update all metrics at minimal cost. However,
since this is risky, we do not try that right now.
Part of bug #12297.
if the undo element we want to add only changes stuff that was already
modified by the previous one on undo stack (in the same group), then
skip it. There is nothing to gain in adding it to the stack.
The typical use case is when doing a search and replace in a large
document that does many replacements in each paragraph. In this case,
the same paragraph would be stored repeatedly.
Fixes bug #12564.
Qt::ImAnchorRectangle has only been introduced in Qt 5.7. Since it is
used to answer a query from the IM machinery, there is no need for
it with older Qt versions.
This was completely broken: the IDs have been escaped in the LyX file
(which they absolutely shouldn't) but not in all LaTeX output (which
they should).
\ was transformed very early to \textbackslash{}, but then the following
routines escaped braces in the string, so we wrongly ended up in
\textbackslash\{\} and "\{} in the output
This is a kind of hack. This allows InsetMathHull to state that it
needs some elbow room beyond its width, in order to fit the numbering
and/or the left margin (with left alignment), which are outside of the
inset itself.
To this end, InsetMathHull::metrics() sets a value in
MetricsInfo::extrawidth and this value is added later to the width of
the row that contains the inset (when this row is tight or shorter
than the max allowed width).
Fixes bug #12320.
The previous commit introduced wrong behaviours for <>. The new code carefully escapes what needs to be escaped from LaTeX, using the now-standard XML tools (XMLStream).
In some cases (I only found numbered equations), the LaTeX output still has a bit of HTML. For numbered equations, using \tag{X} (as suggested in https://www.lyx.org/trac/ticket/13048) would require fiddling with the output. KaTeX doesn't support \label. This patch is probably the easiest and most reliable fix.
Fixes https://www.lyx.org/trac/ticket/13048.
xcode fails with multiple error messages like this one:
----------
CMake Error in po/CMakeLists.txt:
The custom command generating
<build-directory>/po/LyX2.4.cat.pot
is attached to multiple targets:
translations
update-gmo
but none of these is a common dependency of the other(s). This is not
allowed by the Xcode "new build system".
The preview loader assumes all content is in the main document encoding.
As soon as content was not encodable, LyX crashed. We now check for that
and if non-encodable glyphs are found, we do not produce a preview snippet
and warn.
Ideally, the preview loader should be made aware of encoding changes, or
we should generally use utf8 for previews.
The main goal is to match TeX' behaviour. If you want stretchy operators, use InsetMathDelim / \left & \right. This patch will change the output in many files, but making the rendering much closer to that of TeX (which users should expect, due to LyX' roots in TeX).
Book authorship and authorship can be quite distinct. As far as I know, there is no standard way to represent book authorship in DocBook bibliographies.
The new approach to line breaking introduced in 71d9f6e9 is correct,
but the newly introduced min_row_wid should not be updated when the
remaining line width after a break is larger than next_width.
Swapping the two tests fixes remaining issues.
Fixes (for good?) #12899.
The display was wrong when a backing store is in use (which is the case
with Wayland). To fix this in GuiWorkArea::Private::resetScreen(), the
pixelRatio is now stored as a double instead of an int.
Concerning support for QT_SCALE_FACTOR, the existing code was wrong
because this value is already taken into account in devicePixelRatioF
for Qt > 5.6 (no fractional scaling support before that). The
situation is as follows:
Qt < 5.6 5.6 <= Qt < 6 Qt 6
devicePixelRatio int(ratio) int(ratio) ratio
devicePixelRatioF N/A ratio ratio
So it is only between Qt 5.6 and Qt 6 that devicePixelRatioF() has to
be used instead of devicePixelRatio().
QGuiApplication::devicePixelRatio() does not have a 'F' version, it
always returns the real thing.
Fixes ticket #13039.
The issue here is that macro templates are forbidden in InsetTabular
(why? I do not know) and Text::getStatus does not enforce that
properly. Text::insertInset is called and does nothing (because
insertion is forbidden) and yet the cursor is changed to point into
this non existent inset.
Solution:
1/ block insertion of macro templates when not allowed
2/ (additional safety) when insertion of a math macro inset failed, do
not try to set cursor inside the non-existing inset.
Additionally clarify comments.
Before this patch, all table cells had a semicolon at the beginning of their style:
<td style='; border-bottom:
This unwanted behaviour is only due to a typo in a condition when concatenating two styles: the previous code was checking the second part of the style once the first was output, instead of the first one.
In order to perform a conversion of graphic format LyX generates
a python script containing the path of the original file to convert.
While Python2 needs the file in the native encoding of the OS, in
Python3 the encoding has to strictly be how announced by the first
line of a script, in our case utf-8.
Note that there is also a vector parameter that is passed by value.
However it is modified as a local copy and I do not want to change
this now.
Spotted by coverity.
The problem is caused by removing temporary aux, etc, files while
we are in the process of exporting. As Jürgen suggested, we really
do not need to remove these files until we are preparing to export:
They are not used internally. So now we just schedule the removal
and do it then.
colswitch was checking for oldstyle instead of normalcolor
to see if we need \normalcolor, which lead to appearence
of \color{none} in some cases, but "none" is not a known
color.
Instead of displaying the complete path, display the file name
first and then the path enclosed in parenthesis.
In this way files with same name but different path can still be
distinguished and, at the same time, files with different names
but same path are more easily spotted in case the complete path
is truncated by file managers.
While it is not necessary to run validate() on insets that do not
produce output (yellow notes and disabled branches), it has to be done
for previewing, since a construct inside the inset may require some
support.
This is done in two steps:
1. in PreviewLoader::dumpPreamble, indicate that a preview is being
prepared. It is not clear why the `for_preview' boolean was not set
before.
2. in Inset(Branch|Note)::validate, always call the parent's validate
method when previewing.
It should have been possible to move the code from 2. to
InsetText::validate, but the weird code in handling
InsetNoteParams::Comment in html makes it difficult.
It is not clear whether this feature was once documented and used (it
would have been 20 years ago).
It is mostly disable now, but there was a remaining bit that is
annoying now.
I'll try to restore this later.
