The problem with the previous attempt was that, every time through
updateBuffer, we looked up the file location using kpsewhich, which
took too long on Windows. The new solution is to cache that info, and
to look it up only when we need it.
Previously, this info would have been re-read whenever we parsed the
bibfiles. So we re-read it now whenever the bibinfo cache is invalid,
which is less often, but should be good enough. We can add more such
re-reads if need be.
Updates to cmake scripts needed to build with mingw on Fedora.
These are not really right. Things are hardcoded that probably should
not be. These are marked (hopefully) with "# RKH" comments. Possibly
what's needed is another flag? Anyway, help welcome.
On some recent Windows versions, GetLastError() may also return
NO_ERROR instead of ERROR_IO_PENDING during an overlapped write
operation to a pipe. This was confusing the state machine in
Server.cpp so that replies to commands were scheduled but were
never actually output.
The checkProg() function was separating a command from its parameters
by splitting at the first space. This was a problem if the command
was specified with a full path containing spaces. Now the checkProg()
function separates a command from the parameters by splitting at the
first non-quoted space. So, it suffices quoting a path to solve the
issue.
Kornel Benko [Thu, 14 Jun 2018 11:55:52 +0000 (13:55 +0200)]
Cmake build: Adapt to compile with qt5.11
Since qt5_use_modules() is removed from QT5 (as spotted
by Jürgen Spitzmüller), we had to create own version
of this macro.
According to this page:
http://doc.qt.io/qt-5/cmake-manual.html
the way to replace use of qt5_use_modules() for module "_mod" and target "_target"
is to use
1.) cmake_minimum_required(VERSION 3.1.0)
2.) find_package(Qt5${_mod} CONFIG REQUIRED)
3.) target_link_libraries(${_target} Qt5::${_mod})
The last one sets all reguired libraries, compile flags and needed includes for the ${_target}
Disable possible warn about not known policy
Extend Clang compiler detection to cover Apple CLang
Skip paint event when in the middle of a buffer operation
This is detected when an undo group is open and contains at east one
element. This means indeed that changes are in progress. Note that the
group is in general opened in GuiApplication::dispatch. The code there
is changed to ensure that the group is closed before updating the
screen.
This patch is experimental. It is expected to be replaced in master by
a more complete solution. It could in the meantime be backported to 2.3.x.
The problem is that popping dialogs during reload can cause paint
events for which we are not ready. If this does not work, then we
can introduce a new flag, besides 'busy', for that case. But busy
does not seem to be used very widely, so hopefully this works.
Note that when paintEvent is triggered on macOS, the least we can do
is to copy our backing store to the screen.
Completion uses a complicated async scheme based on timers to update
itself. This is probably not necessary anymore and is fragile to
event order changes.
This is what happens with the new painting scheme. Therefore the
asyncHideXXX() methods have to be made more robust and detect whether
completion state has changed by the time they are triggered.
When an image has to be loaded a second time and the cache
is not used or the image is not in the cache, its preview is
going to be regenerated again. Hence, if the same image
appears more than once on screen, avoid requesting simultaneous
generation of the same preview.
See also this thread:
https://www.mail-archive.com/lyx-devel@lists.lyx.org/msg205057.html
Set caret height correctly for cells inside math rows
The code in 90cfe4ec3 only handled the cells which metrics are
computed directly, and missed those who were linearized inside a MathRow.
To fix this, we use the fact that all the positions in a math row have
the same height and make MathRow::metrics return a boolean indicating
whether it contains a caret for a given bufferview.
Stephan Witt [Sun, 13 May 2018 18:15:35 +0000 (20:15 +0200)]
#11142 correct list of previous versions to check for user directory contents
LyX on Mac uses a user directory with version suffix. On change of the version suffix
the existence of the directories with previous versions is checked and the latest one
is used for a copy on first configure run.
For 2.3 the candidate list starts with 2.2 now as it should.
Compute a height from current font and current cell vertical
dimensions in MathData::metrics(), because this is where current font
is known.
Introduce BufferView::setCaretAscentDescent to remember this value.
This mechanism is not used for text because Cursor::current_font is
restored by undo, and the caret height would not be changed then. But
in principle it is doable.
When computing a cell metrics, it is now possible to specify whether it
is tight (at least as tall as 'x') or not (as tall as the max height of
the font).
Use this to make sure that grid insets have large enough cells. It
will probably appear that other cells needn't be tight. Currently, the
only cell which is known to be tight is the nucleus of the root inset.
Others should be examined one by one. It might be that the default of
MathData::metrics tight parameter should be `false'.
Scott Kostyshak [Mon, 30 Apr 2018 00:06:16 +0000 (20:06 -0400)]
Do not use needauth when only exporting code
We have a couple of converters (using Sweave and knitr to "tangle"
an intermediate file) that are used for exporting code chunks
contained in a .lyx file. Since the code is just exported to a text
file and is not executed, needauth is not necessary.
It surprises me a bit that the code is different in this case
from the case when EXPORT_in_THREAD is not set. I should check
to see if there's a problem in the other case, too.