Revert last mods in TGListTree and do it in a different way - remove Cocoa specific ifdefs from DrawItemName, but also put ytext modification in ifdefs (when it's adjusted using pic sizes).
Still, outline and highlight are painted with full item's height.


git-svn-id: http://root.cern.ch/svn/root/trunk@47075 27541ba8-7e3a-0410-8455-c3a389f83636

to use font height in different DrawXXX methods + I have to use the full item height,
otherwise (as soon as FontAscent is smaller then FontHeight and smaller than image(s) height,
outline, highlight, text - everything is misplaced).


git-svn-id: http://root.cern.ch/svn/root/trunk@46959 27541ba8-7e3a-0410-8455-c3a389f83636

1. When it comes to tiny polygons and pixel-exact rendering, Quartz does not produce expected image
   (well, it does with HiDPI, but HiDPI has additional problems). So I have to adjust small triangles in 
   a color selector and pattern selector widgets.
2. It's really bad idea to draw a rectangle first and then try to fill some area inside this rectangle,
   trying not to overlap - this does not work with Quartz, so I have to invert the order: fill rectangle first
   and draw outline later.

These fixes require ifdefs, ugly but inevitable, if you do some hardcoded pixel-based rendering as we do.


git-svn-id: http://root.cern.ch/svn/root/trunk@46934 27541ba8-7e3a-0410-8455-c3a389f83636

Sizes are hardcoded in a GUI, layout manager ignores the fact widgets do not fit (may be, it should not), but
as a result on Cocoa text label "Draw option:" is overlapped by button in a TBrowser with default geometry.
I have to ask for these two pixels back.


git-svn-id: http://root.cern.ch/svn/root/trunk@46929 27541ba8-7e3a-0410-8455-c3a389f83636

created as first ROOT object. This happens only when statically linking
the application.


git-svn-id: http://root.cern.ch/svn/root/trunk@46928 27541ba8-7e3a-0410-8455-c3a389f83636

Fix savannah bug #98532: TGuiBuilder: crash when changing text of a field (https://savannah.cern.ch/bugs/?98532)


git-svn-id: http://root.cern.ch/svn/root/trunk@46888 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@46849 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@46119 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@46073 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@45666 27541ba8-7e3a-0410-8455-c3a389f83636

- move enum ENewBrowserMessages into TRootBrowser.h class definition
- make CloseTab() virtual
- make CloseWindow() virtual


git-svn-id: http://root.cern.ch/svn/root/trunk@45593 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@45565 27541ba8-7e3a-0410-8455-c3a389f83636

Fix for right mouse buttons.


git-svn-id: http://root.cern.ch/svn/root/trunk@45547 27541ba8-7e3a-0410-8455-c3a389f83636

Modification for different popups: with cocoa I can not guarantee the stack order for borderless popup window
(it's "main" window can in some cases become main and key window and overlay popup).
With transient hint (which makes a popup window into a children of some other window) the problem is fixed,
now GUI does not fall to pieces at any unexpected user action.


git-svn-id: http://root.cern.ch/svn/root/trunk@45252 27541ba8-7e3a-0410-8455-c3a389f83636

window->HandleEvent, so it wasn't required.


git-svn-id: http://root.cern.ch/svn/root/trunk@45158 27541ba8-7e3a-0410-8455-c3a389f83636

Draw triangle with right (black) GC when the pattern select is enabled (thanks Timur for reporting it)


git-svn-id: http://root.cern.ch/svn/root/trunk@45146 27541ba8-7e3a-0410-8455-c3a389f83636

Draw triangle with right (black) GC when the color select is enabled (thanks Timur for reporting it)


git-svn-id: http://root.cern.ch/svn/root/trunk@45145 27541ba8-7e3a-0410-8455-c3a389f83636

process "paint events" immediately, but ROOT also can draw outside of paint event.
And with combobox there are 2 "rendering passes", one is with incorrect sizes, the second one 
is with correct sizes. With X11/Win32 these passes are sequenced - and the second
one is initiated by expose event. With Cocoa the second pass actually happens before the 
first one. This, plus some implementation details and different event order results
in quite ugly visual artefact - the button in the right part of combobox is 
not correctly repainted.


git-svn-id: http://root.cern.ch/svn/root/trunk@45132 27541ba8-7e3a-0410-8455-c3a389f83636

2. Remove modifications from TRootCanvas (accident).


git-svn-id: http://root.cern.ch/svn/root/trunk@45007 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44997 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44996 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44899 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44773 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44708 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44687 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44686 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44685 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44658 27541ba8-7e3a-0410-8455-c3a389f83636

  using namespace std;
from Riostream.h, which has huge consequences for all of ROOT.
Riostream.h is now a simple wrapper for fstream, iostream, iomanip for backward compatibility; Riosfwd.h simply wraps iosfwd.

Because of templates and their inline functions, Riostream.h needed to be included in headers, too (e.g. TParameter.h), which violated the assumption that Riostream.h is not exposing its using namespace std to headers.
ROOT now requires R__ANSISTREAM, R__SSTREAM, which does not change the set of supported compilers.

Without "using namespace std", several identifiers are now prefixed by std::; e.g. roofit/* source files now have a using namespace std to keep their coding style.
TFile::MakeProject() now generates "using namespace std" to convert the CINT-style class names into C++ ones.


git-svn-id: http://root.cern.ch/svn/root/trunk@44507 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44195 27541ba8-7e3a-0410-8455-c3a389f83636

   thus constant (by nature) function modifies object's state. This makes it impossible
   to paint the same widget twice by calling DoRedraw, the second time exposed region is empty.
   This does not work with cocoa.
2. CreateFontCollection for a specific name always finds only 1 font, not all variations of the same font,
   instead I have to request the full list every time and filter it myself.
3. TGFontDialog: connects signal/slot only for gVirtualX inherited from TGX11 (or TGX11 itself),
   must also work for TGCocoa. all_sizes and all_styles variables also were set incorrectly for TGCocoa
   (initialization uses check InheritsFrom("TGX11")).


git-svn-id: http://root.cern.ch/svn/root/trunk@44185 27541ba8-7e3a-0410-8455-c3a389f83636

Fix to prevent a potential crash when deleting canvas with toolbar, as reported on the forum (http://root.cern.ch/phpBB3//viewtopic.php?f=3&t=14627)


git-svn-id: http://root.cern.ch/svn/root/trunk@44173 27541ba8-7e3a-0410-8455-c3a389f83636

color wheel object creates a canvas (step 1), appends itself 
into this canvas (step 2). This code relies on the fact, that
all events (configure/expose/etc) will be received _after_ step 2
and canvas will be eventually repainted later. This is not true in case
of cocoa: window is created and made visible on step 1, it's repainted
after step 1 and before step 2, so any contents added after step 2
will become visible only after gPad->Update() call (BTW, this is how it's 
done in TRootCanvas for "View"->"Markers".

Call gPad->Update(), do not rely on X11 specific (events in a queue at the end
of event loop iteration).


git-svn-id: http://root.cern.ch/svn/root/trunk@44170 27541ba8-7e3a-0410-8455-c3a389f83636

it crashes - parses this XLFD, sets foundry to 0 and later use strcasecmp on this pointer.
1. 'Fix' TGFontPool not to use 0 pointers,
2. return fake foundry.


git-svn-id: http://root.cern.ch/svn/root/trunk@44151 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44149 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44086 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44076 27541ba8-7e3a-0410-8455-c3a389f83636

TGView frame (base of TGTextView) has a child window - TGViewFrame.
TGViewFrame is never actually painted (DoRedraw never gets called for this window),
but ... it receives Expose event and delegates it to ... its parent - TGTextView,
and TGTextView, it its DoRedraw (or DrawRegion) .... tries to paint into child window - TGViewFrame.
While this is may be OK for X11 or Window, with Cocoa it's a total mess and disaster - if some view is
to be updated, it does not expect a child to be suddenly rendered in the middle of work.

View is not a window.

I've added some quite risky code to address this problem (text scrolling is still wrong) - only 
"static rendering" for a text view works now.


git-svn-id: http://root.cern.ch/svn/root/trunk@44075 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@44062 27541ba8-7e3a-0410-8455-c3a389f83636

git-svn-id: http://root.cern.ch/svn/root/trunk@43986 27541ba8-7e3a-0410-8455-c3a389f83636