The window title is built from the current file name and its
mofidication state. We use our own code instead of the automatic title
bar provided when windowFileName() is set because
1/ Qt does not keep the full path name
2/ Qt does not yield a nice application name
3/ Qt separates file name and app name with an em-dash. It seems that
only KDE does that: Gnome does not, Windows does not either. I do not
think that we can/want to detect a KDE environment at run-time.
The "read only" and "version control" status are shown in the status bar:
* for read only we use the tab read only emblem (with the right size)
* for version control, we show the name of the backend (using a new
vcname() method of the backend).
The iconText() of the view is not updated anymore, since this is
deprecated in Qt5.
(cherry picked from commit 82808fea04)
(cherry picked from commit 9313f8565b)
(cherry picked from commit bbe36ce6ce)
(cherry picked from commit 7cbc0c16a8)
Do not assume that the /systemlyxdir path prefix in \origin refers
to the system directory of the running instance, but check through
some heuristics what the real system dir is. In this way, a document
in the system dir of any other LyX installation is correctly spotted
and the \origin tag properly updated. For example, one can use an
installed version of lyx to edit a document in the lib/doc dir of a
git repo and obtain the same result as when running lyx in place.
(cherry picked from commit ade098e289)
If the first character in the first cell of an aligned math environment is
'[', and the environment does not use top or bottom vertical alignment,
then LyX did write the '[' unprotected so that it got misinterpreted as
optional argument, both when reading the .lyx file in LyX and when reading
the .tex file in LaTeX => data loss!
The fix is to output an empty optional argument in this case, which is
interpreted as default alignment both by LyX and LaTeX. It would also be
possible to output \[ in the first cell instead, but this would be more
difficult to implement.
This is a fixup to commit 39329935. The two fixes are
* add forgotten offset `y' when drawing the line
* in order to have a continuous vertical line, draw from the offset of
the previous row.
Fixes bug #10363.
(cherry-picked from a27ff136)
We do already have docx, but xlsx was missing. This is a separate format
because of the MIME type. nd because some users might need converters which
can only handle one format. Now the spreadsheet template does not hide the
fact anymore that it can deal with xlsx files as well.
When the Qt directory is specified by --with-qt-dir, the tools
are only searched for in the corresponding bin subdir and their
full path is retained, so that it is not necessary modifying the
PATH and PKG_CONFIG_PATH environment variables to select a custom
Qt version. Otherwise, if qtchooser is detected early in the PATH,
the generic names are attempted with proper arguments to select the
desired Qt version (e.g., "moc -qt=qt5" if --enable-qt5 is specified).
Failing all of the above, the generic names with and without proper
suffixes (either -qt4 or -qt5) are checked in the PATH.
Eventually, a check is performed and a warning is issued if the
selected tools do not match the chosen Qt libraries.
This removes the submenu indirection in Insert and the type changer in
contextual menu. Interestingly, the code was there, but it did not
work at all.
(cherry picked from commit e7a33cacf1
and made compatible with C++98)
When breaking paragraph in an empty top-level paragraph, nothing
happens on screen but yet there is an undo step because the layout is
reset to what it already was.
Avoid this case.
Fixes bug #10089.
(cherry picked from commit 2ad52918da)
Syntax highlighting now provides the appropriate cue that the user preamble is
inside \makeatletter…\makeatother.
(cherry picked from commit 10f6eb2e7e)
On Windows, runCommand has never returned the exit code of the
spawned process but the result of correctly closing the input handle.
Fixes#10327
(cherry picked from commit fb46b3a1ee)
There was an oddity in the manual that exposed a problem with the
test for the "special case" of an inset all by itself in a pargraph.
If a font change is applied to that inset, we still need to open the
paragraph.
(cherry picked from commit 3be23d18b6)
The export of compressed files to previous lyx versions failed on windows.
The change in this case is to explicitly pass the output file since we know it.
At the same time take care of the fact that lyx2lyx knows what file format
versions corresponds to each lyx version.
care of much of what we need to do for RTL languages. It does not
take care of inline language changes, probably.
(cherry picked from commit 07dcb1c525)
dealing properly with the paragraph separator tag.
We really need to use that tag as a kind of general marker for which
tags we're responsible for in a given paragraph and which tags we are
not. So the changes to InsetText.cpp use the tag as that kind of marker.
Note that, as of this commit, the User Guide again exports without any
kind of error. I haven't yet checked the other manuals.
This fixes bug #8022.
(cherry picked from commit 31e25c8ec6)
In redoParagraph, this should be done before coping with the insets,
other wise some graphic gliches may occur. This is a better fix for
Fixes bug #10163.
(cherry picked from commit 71374b38c2)
LyX did not distinguish compressed and uncompressed svg files previously.
Therefore XHTML export of vector graphics did use svgz images directly, which
is not supported by browsers. If svg and svgz are treated as two formats then
all works fine. This is also consistent with the loadable image formats
reported by qt: It reports both svg and svgz.
The gunzip dependency in converters is not new (it is already used internally),
but the gzip dependency is new, so it might not be available on windows.
This is not important at the moment, since we do not yet need to convert svg
to svgz, I only added the converter for completeness.
This is the well known file locking problem: The TempFile class keeps the
created file locked for the own process, and this prevents the CAS to read it.
Instead of resetting it to false, do a proper test to see whether
there is a separator at the end of the row.
Fixes bug #10180.
(cherry picked from commit 5e5440f2f3)
