Files not explicitly opened in binary mode are text files and thus
expect a 'str' arg for their write() method.
Python 2 is nearing its EOL and e.g. Debian has already
removed the python2-based 'python-polib' package previously
used here from its testing distribution.
I checked that running the commands
./scripts/unocommands.py --update . ../libreoffice
./scripts/unocommands.py --translate . ../libreoffice/translations
./scripts/unocommands.py --check .
still yields the same results as previously.
Change-Id: I39e1785d3c78416009420dd4c2be58bd1c3647c3
Reviewed-on: https://gerrit.libreoffice.org/81422
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Michael Stahl <michael.stahl@cib.de>
When the user selects an item (touch), the LO server updates and send
selected data to apply border styles.
Change-Id: Ic6f877712514806b822d70342738ee04ae2ea0a8
Reviewed-on: https://gerrit.libreoffice.org/81419
Reviewed-by: Henry Castro <hcastro@collabora.com>
Tested-by: Henry Castro <hcastro@collabora.com>
sidebar of the impress document
The client side sends the JSON string:
{
id: "sd::LayoutMenu", // control id in sidebar LayoutMenu
cmd: "selected", // Selection action
data: "3" // item id,
}
The LO core server receives the data and execute actions to the
correspond control child.
Change-Id: I066b804a63f48fe04e9c691da928e186b573f8c2
Reviewed-on: https://gerrit.libreoffice.org/81168
Reviewed-by: Henry Castro <hcastro@collabora.com>
Tested-by: Henry Castro <hcastro@collabora.com>
74ce4a9261 introduced a
different order than in core.
Change-Id: I4d601e1fcd3028c7bbe9d0119c15193fe1db1f35
Reviewed-on: https://gerrit.libreoffice.org/81270
Reviewed-by: Aron Budea <aron.budea@collabora.com>
Tested-by: Aron Budea <aron.budea@collabora.com>
This patch fixes the following problems:
1)
- create a document with several comments
- pan so that the comments are visible and the commented text is not
- tap on a comment box: no scroll action is performed for getting the
commented text visible
2)
Writer: in a document with a single page and a few comments,
clicking/tapping on a comment box doesn't align it with the commented
text since it is not possible to scroll the document (it is already
fully visible - it's made by a single page) and the comments are not
scrolled up/down.
Change-Id: Iaedbe9a89b5f45c322c5f9405960017cd7054e8e
This patch mitigates but does not solve completely a problem that
seems to affect especially iOS devices when the user performs a pinch
to zoom the comments layout is not updated.
Change-Id: I04a0e91c11d07744653b6a25ffb553d22249bcb3
The selected comment can be placed far away from not selected comments
so the intersection could be empty, better to check for the y
coordinate only.
Change-Id: I2f649521bc5de44c89cbb61f76e6930aa5ccfe96
extraSize is fixed because the comments are never
scaled themselves, so we shouldn't scale the extraSize.
This fixes the issue with not being able to see the
comment boxes fully, even when scrolling to the far right,
when the screen width is too small.
What happened with small screens is that the document
zoom factor (and hence the dimensions scaling) kicked
in and zoomed the document out (< 1 scaling factor).
So when we scaled the extraSize needed by the comments,
we reduced their available width and since the
comment boxes aren't scaled they were left outside
the document and outside the reach of the user.
Change-Id: I57df55963b6119d5dde878bf3e1c3ffe984b7c43
* map may not have hasLayer defined during
initialization, so we protect against it.
* Show the inner html elements of the comment
boxes before showing the comment itself (smoother
UI experience).
* Reorder getBounds call first to short-circuit
early.
Change-Id: I0235d43d19c1b712df266d6e39738b1415c5d048
The ieee 754 supports multiple representations
for the same mathematical value (such as 0 and -0).
Worse, because not all numbers can be represented
exactly, mathematically identical numbers will
not be equal (such as 0.1+0.2 != 0.3).
The proper way to compare floating-point numbers
is, therefore, to use the epsilon.
It was observed through debugging that often
we triggered layouting of comments even though
their coordinates hadn't changed, because the naive
!= comparison returned true. Also was observed
that when checking for (0,0) coordinates, sometimes
our value would be -0, which again the naive ==
comparison yields false.
Change-Id: I246fe02d3db4132bd1d7e7a1761d8a9ae1686fb2
Call focus on the input field of the comment dialog after everything is
set. This focus call is also made by vex.dialog.open() method, but
the focus is lost by the modifications done after the open method.
Mobile only.
Change-Id: I6a8b9105a0682141d9cad12e3aeb41fd0efeecc5
scroll up to 200, enter some text, add a new sheet, headers, grid
lines are wrong, swicth back to the previous sheet, now headers and
document position are wrong too
Change-Id: If9d59b5509584e92491bbce4b8db54ec5308b211
The wrong index prevented the execution of the code in Map.setPart
since `part < docLayer._parts` check failed.
Change-Id: I3077ffb68decc464e19ff7d54468ba5582bdc21e
That has been done in order to avoid to trigger the context menu too
early; so the user can start panning in a relaxed way.
Change-Id: I64657efd55649584be74c96a31fbe2199a25f91a
This patch handles a new flag attached to the invalidate view cursor
message for informing the client when the text cursor is inside a
mispelled word.
This information is used for popping up the spelling context menu on a
long press event instead of the standard context menu for a selected
word.
Change-Id: Idf5994c022d66369f05f13b8549994155c9fad18
Selection handlers dragging is not handled through hammer so the event
handlers registered for hammer need to be disabled on dragging start
and enabled on dragging end: this is done by invoking
TouchGesture.removeHooks and TouchGesture.addHooks, anyway
TouchGesture.removeHooks didn't remove the 'press' event, while
TouchGesture.addHooks was adding a new 'press' event handler every
time the selection handler dragging ended.
Change-Id: I557beae1ac66fd2095b996e08dcf2eac611d216a
for selecting text we need to simulate a double click, anyway for a
graphic object a single click is enough, while a double click can lead
to switch to edit mode (not only for an embedded ole object, even for
entering text inside a shape);
a similar problem regards spreadsheet cell: a single click moves the
cell cursor, while a double click enables text input;
in order to avoid these cases, we send a single click and wait for a
few milliseconds before checking if we received a possible selection
message; if no such message is received we simulate a double click for
trying to select text and finally, in any case, we trigger the context
menu by sending a right click;
Change-Id: I16722b5f1208b9bb22c613a8417c9c34ac310033
sometimes online receives several messages from core for the same
context menu: that probably leaves the jquery context-menu in a dirty
state, causing to not update it correctly with the new menu entries
Change-Id: I45356670978a8e8eb2c1ef5e32276b08d5042c41
So in same case pinch event is triggered even on a single tap. In that
cases the _onPinchEnd animation fails causing to not change
cursors/handlers opacity back to 1, for that reason I pushed a patch
which move the animation stuff to the end of _onPinchEnd and moreover
I even check that the center parameter is really defined.
Change-Id: I1f1c98fb3ad056aa14ee5657e2f5f1cf489789eb
