wxWidgets debian packages for Karmic Koala
projek...@astade.de
Hello,
I'm maintainer of the Astade UML Tool Project http://astade.de.
For our project we have an Ubuntu repository, from where you can install
the tool: http://apt.astade.de.
Astade needs wxWidgets 2.8 ansi build
In the past I took the preecompiled debian packages from apt.wxwidgets.org
and copy them into the Astade repository at apt.astade.de
Now I want to add a "Karmic Koala" section to the Astade repository.
For that I need the wxWidgets 2.8 ansi debian package compiled for
"Karmic Koala".
Unfortunately there is none in apt.wxwidgets.org :-(
Do you have any ideas, where to get it?
I already tried, unsuccessful, to build it myself. Maybe I could manage this
if I spend some more time. But even than, I would prefer an "official"
version.
Thanks in advance.
P.S.: wxWidgets 2.8 is in the official Ubuntu repositories now. But
only the unicode build, as far as I know :-(
Robin Dunn
>
>
> Hello,
>
> I'm maintainer of the Astade UML Tool Project http://astade.de.
> For our project we have an Ubuntu repository, from where you can install
> the tool: http://apt.astade.de.
>
> Astade needs wxWidgets 2.8 ansi build
>
> In the past I took the preecompiled debian packages from apt.wxwidgets.org
> and copy them into the Astade repository at apt.astade.de
>
> Now I want to add a "Karmic Koala" section to the Astade repository.
> For that I need the wxWidgets 2.8 ansi debian package compiled for
> "Karmic Koala".
>
> Unfortunately there is none in apt.wxwidgets.org :-(
>
> Do you have any ideas, where to get it?
Do the packages for Jaunty work?
--
Robin Dunn
Software Craftsman
http://wxPython.org
Carlos Rodriguez
I'm using Jaunty packages on Karmic and it works without problem.
Robin Dunn escribió:
--
========================
Carlos Rodriguez Llodrá
carlos....@gmail.com
Thomas Spitzer
>
> I'm using Jaunty packages on Karmic and it works without problem.
"Just use the Jaunty packags" is the message? Ok I'll do that.
I'd expected some mystic build secret, not just "lets try wether the
Jaunty packages work". But it's fine for me.
Obviously I think to complicate ;-)
Thanks
Thomas
Thomas Spitzer
> I'd expected some mystic build secret, not just "lets try wether the
> Jaunty packages work". But it's fine for me.
> Obviously I think to complicate ;-)
OK, I made the test. First I installed the whole package from "deb
http://apt.astade.de jaunty main" at my desktop PC which is a Jaunty
system. As expected, everything runs fine.
Next I installed a "Karmic Koala" in a "Sun virtual Box" virtual
machine, running on my Jaunty desktop. I added the Jaunty repository to
the Karmic system and installed the package. It runs not really bad. But
when I start "AstadeDraw" add an Actor and an Usecase and try to
associate the Actor to the Usecase, it is not working. Just no
connection cursor appeares. Remember: This works in Jaunty!
Actually it is possible, that it is because of the virtual machine?
But at this state of the test it looks, as if the Jaunty packages are
not running perfectly correct in Karmic.
What next? Test in a Jaunty virtual machine?
That'll take again some time.
Carlos Rodriguez
in Karmic, because it was doing some really weird things with user input
on *some* applications (not wx-related, but...)
On the other hand, I'm downloading AstadeDraw right now to test that ;-)
Thomas Spitzer escribió:
> Please read http://www.wxwidgets.org/support/mlhowto.htm before posting.
>
> To unsubscribe, send email to wx-users+u...@googlegroups.com
> or visit http://groups.google.com/group/wx-users
> -~----------~----~----~----~------~----~------~--~---
Thomas Spitzer
> that ;-)
Just to be sure, here is a short descroption:
1. Install as described here:
http://astade.de/doku0c34.html?id=install:installinubuntu
2. Start AstadeDraw
3. Rightclick somewhere inside the window and select "new Actor" in the
context menu (You get an Actor, which is floating towards the middle of
the window)
4. Rightclick again and select "new usecase" you get an "egg" which
floats away, from the actor.
5. Rightclick the actor and select "Association to..." when you move the
mouse, you should get a dotted line, from mouse position to the actor.
(see attached graphic)
In my test the dotted line was not there and it was not possible to
attach to the usecase. In Jaunty it is, Compiz is enabled!
Greetings
Thomas
Carlos
But, I've not had any trouble with other wx applications (audacity, code::blocks), although they're not using exactly the same libraries (unicode vs ansi).
--
Thomas Spitzer
> also after updating everything and still nothing :-(
>
> But, I've not had any trouble with other wx applications (audacity,
> code::blocks), although they're not using exactly the same libraries
> (unicode vs ansi).
Is somebody able to build the "official" karmic debian packages?
Vadim Zeitlin
TS> > I've tried it in a clean Karmic i386 install, and didn't work... tried
TS> > also after updating everything and still nothing :-(
TS> >
TS> > But, I've not had any trouble with other wx applications (audacity,
TS> > code::blocks), although they're not using exactly the same libraries
TS> > (unicode vs ansi).
TS>
TS> Which brings me back to the initial question:
TS> Is somebody able to build the "official" karmic debian packages?
I think your best bet would be to build them yourself using the
instructions in debian/README.HowToBuild. Normally it should go without
problems...
Regards,
VZ
--
TT-Solutions: wxWidgets consultancy and technical support
http://www.tt-solutions.com/
Thomas Spitzer
> instructions in debian/README.HowToBuild. Normally it should go without
> problems...
Thomas Spitzer
first on my Jaunty desktop.
It seams, that I've not yet managed to get the correct source tree.
First I did:
svn co http://svn.wxwidgets.org/svn/wx/wxWidgets/tags/WX_2_8_10/
But "debian/build_all" complains:
fakeroot debian/rules clean
cd: 1: can't cd to wxPython
make: *** [clean-py] Fehler 2
dpkg-buildpackage: Fehlschlag: fakeroot debian/rules clean gab
Fehler-Exitstatus 2
It is right, there is no directory named "wxPython"
So I did another try:
svn co http://svn.wxwidgets.org/svn/wx/wxPython/tags/wxPy-2.8.10.1
Here "debian/build_all" complains:
mkdir: kann Verzeichnis „./_dist_dir/wxGTK-2.8.10/src/msw“ nicht
anlegen: File exists
make: *** [debian-msw-dirs] Fehler 1
So for me it looks, that I need a correct source tree from somewhere.
But at the moment I've no glue, where to get the "correct" one.
Robin Dunn
then probably the easiest approach to take is to use the source debs.
Something like this:
* With the apt.wxwidgets.org jaunty repository in your
/etc/apt/sources.list (see
http://wiki.wxpython.org/InstallingOnUbuntuOrDebian) do:
apt-get source -d wxwidgets2.8
* Unpack the source files and prepare it for building:
dpkg-source -x wxwidgets2.8_2.8.10.1-1.dsc
* Build the binary packages:
cd wxwidgets2.8-2.8.10.1/
fakeroot debian/rules binary
Thomas Spitzer
> Since you are wanting to build packages for an already released version
> then probably the easiest approach to take is to use the source debs.
> Something like this:
>
> * With the apt.wxwidgets.org jaunty repository in your
> /etc/apt/sources.list (see
> http://wiki.wxpython.org/InstallingOnUbuntuOrDebian) do:
>
> apt-get source -d wxwidgets2.8
>
> * Unpack the source files and prepare it for building:
>
> dpkg-source -x wxwidgets2.8_2.8.10.1-1.dsc
>
> * Build the binary packages:
>
> cd wxwidgets2.8-2.8.10.1/
> fakeroot debian/rules binary
>
The 4 commands above produced (for jaunty):
libwxbase2.8-0_2.8.9.1-0ubuntu6_i386.deb
libwxbase2.8-dbg_2.8.9.1-0ubuntu6_i386.deb
libwxbase2.8-dev_2.8.9.1-0ubuntu6_i386.deb
libwxgtk2.8-0_2.8.9.1-0ubuntu6_i386.deb
libwxgtk2.8-dbg_2.8.9.1-0ubuntu6_i386.deb
libwxgtk2.8-dev_2.8.9.1-0ubuntu6_i386.deb
python-wxaddons_2.8.9.1-0ubuntu6_all.deb
python-wxgtk2.8_2.8.9.1-0ubuntu6_i386.deb
python-wxgtk2.8-dbg_2.8.9.1-0ubuntu6_i386.deb
python-wxtools_2.8.9.1-0ubuntu6_all.deb
python-wxversion_2.8.9.1-0ubuntu6_all.deb
wx2.8-doc_2.8.9.1-0ubuntu6_all.deb
wx2.8-examples_2.8.9.1-0ubuntu6_all.deb
wx2.8-headers_2.8.9.1-0ubuntu6_i386.deb
wx2.8-i18n_2.8.9.1-0ubuntu6_all.deb
wx-common_2.8.9.1-0ubuntu6_i386.deb
but if you look at
http://apt.wxwidgets.org/dists/jaunty-wx/main/binary-i386/ you see, that
I need the two packages named:
libwxbase2.8-0-ansi_2.8.10.1-0_i386.deb
libwxgtk2.8-0-ansi_2.8.10.1-0_i386.deb
Maybe you can tell me, how to get them? Even if I rename the corespnding
packages from above (may I simply rename them?) they are still not ansi
packages, are they?
best regards
Thomas
Vadim Zeitlin
TS> but if you look at
TS> http://apt.wxwidgets.org/dists/jaunty-wx/main/binary-i386/ you see, that
TS> I need the two packages named:
TS>
TS> libwxbase2.8-0-ansi_2.8.10.1-0_i386.deb
TS> libwxgtk2.8-0-ansi_2.8.10.1-0_i386.deb
TS>
TS> Maybe you can tell me, how to get them?
You need to set WX_UNICODE to 0 in debian/rules.
Thomas Spitzer
I build in "Karmic" now.
It runs pretty far. than I get this error:
cd wxPython \
&& python2.5-dbg ./setup.py
build \
WX_CONFIG='/home/thomas/wxBuild/wxwidgets2.8-2.8.10.1/objs_gtk_d/wx-config --no_rpath' \
WXPORT=gtk2 \
\
FLAVOUR=dbg
WARNING: WXWIN not set in environment. Assuming '..'
Fatal Python error: Interpreter not initialized (version mismatch?)
Aborted
Maybe it has to do with the fact, that it explicitely uses python2.5 and
"Karmic" per default installs python2.6 ?
Or what else is the problem here?
Regards
Thomas
Vadim Zeitlin
TS> > You need to set WX_UNICODE to 0 in debian/rules.
TS>
TS> OK, things work much better, now.
TS> I build in "Karmic" now.
TS> It runs pretty far. than I get this error:
TS>
TS> cd wxPython \
TS> && python2.5-dbg ./setup.py
TS> build \
TS>
TS> WX_CONFIG='/home/thomas/wxBuild/wxwidgets2.8-2.8.10.1/objs_gtk_d/wx-config --no_rpath' \
TS> WXPORT=gtk2 \
TS> \
TS> FLAVOUR=dbg
TS> WARNING: WXWIN not set in environment. Assuming '..'
TS> Fatal Python error: Interpreter not initialized (version mismatch?)
TS> Aborted
TS>
TS>
TS> Maybe it has to do with the fact, that it explicitely uses python2.5 and
TS> "Karmic" per default installs python2.6 ?
This looks the likely cause of the problem, yes. I think wxPython should
work with 2.6 too. OTOH if you're not interested in wxPython you could just
skip building it altogether too.
Thomas Spitzer
> work with 2.6 too. OTOH if you're not interested in wxPython you could just
> skip building it altogether too.
I've placed them to the repository astade.homeip.net/apt
(which is my experimental repository) and installed Astade in Karmic.
Unfortunately there is still the bug.
AstadeDraw is not running correct.
What should I do now?
Regards
Thomas
Vadim Zeitlin
TS> > This looks the likely cause of the problem, yes. I think wxPython should
TS> > work with 2.6 too. OTOH if you're not interested in wxPython you could just
TS> > skip building it altogether too.
TS>
TS> So, now I've managed to build the packages.
TS> I've placed them to the repository astade.homeip.net/apt
TS> (which is my experimental repository) and installed Astade in Karmic.
TS>
TS> Unfortunately there is still the bug.
TS> AstadeDraw is not running correct.
TS>
TS> What should I do now?
Unfortunately the description of the problem is so vague that I don't even
really know where to start... What exactly is not correct?
Also, does it have anything to do with packages at all? I.e. does it work
correctly if you forget about packaging and just use wxWidgets in the
"usual" way?
Thomas Spitzer
> really know where to start... What exactly is not correct?
>
> Also, does it have anything to do with packages at all? I.e. does it work
> correctly if you forget about packaging and just use wxWidgets in the
> "usual" way?
system. The Code needs wxWidgets in an ansi build.
If I take the Jaunty build binaries and run them together with the
jaunty wxWidgets packages from apt.wxwidgets.org in a jaunty system,
everything runs fine.
Next I tried to run the same packages in Karmic. Same Astade binary,
same wxPackages. It looks not so bad, at the first glance, but when
running AstadeDraw there is an error.
Error description: AstadeDraw has a window, where you can create nodes,
by rightclick and select from a context menu. You might connect this
nodes by rightclick a node and select e.g. "Associate to ..".
moving the mouse to the other node, you get a "dotted line" and when you
click at the second node, the connection is made (look attached
picture). In Karmic you might create nodes, but when you try to connect
them, you do not get the dotted line, and when you select the second
node anyway, there is no connection.
Assumption: I need Karmic wxWidgets packages. That was the reason I
tried to build them. But after having them, now, I discovered, the error
is still there.
Now the question: what to do next? First idea: debugging!
but than I've to be able to build in karmic. Than I need the development
packages?
Second idea: I make a tracing version of AstadeDraw which traces all
function calls. I run this in Jaunty and in Karmic and compare the two
traces. Than we have an idea, where the bug comes from. That'll take
some time. If I've done that, I'll post the result :-(
Regards
Thomas
Vadim Zeitlin
TS> Next I tried to run the same packages in Karmic. Same Astade binary,
TS> same wxPackages. It looks not so bad, at the first glance, but when
TS> running AstadeDraw there is an error.
TS>
TS> Error description: AstadeDraw has a window, where you can create nodes,
TS> by rightclick and select from a context menu. You might connect this
TS> nodes by rightclick a node and select e.g. "Associate to ..".
TS> moving the mouse to the other node, you get a "dotted line" and when you
TS> click at the second node, the connection is made (look attached
TS> picture). In Karmic you might create nodes, but when you try to connect
TS> them, you do not get the dotted line, and when you select the second
TS> node anyway, there is no connection.
I think this problem could be due to the change in GTK+ version. AFAIR
Karmic uses GTK+ 2.18 which has many changes compared to the previous
versions and one of them might have broken wxGTK drawing code. This is pure
guesswork, of course.
TS> Now the question: what to do next? First idea: debugging!
TS> but than I've to be able to build in karmic. Than I need the development
TS> packages?
You don't really need packages at all, you can just get wx sources from a
tarball or svn and build them with --enable-debug.
TS> Second idea: I make a tracing version of AstadeDraw which traces all
TS> function calls. I run this in Jaunty and in Karmic and compare the two
TS> traces. Than we have an idea, where the bug comes from. That'll take
TS> some time. If I've done that, I'll post the result :-(
This could be useful but I suspect that there won't be any difference
between the traces. This really does look like a problem with wxGTK
refresh/redrawing and latest GTK+. Unfortunately I don't know what could it
be exactly.
One proposal I do have is for you to try 2.9.0 or the latest svn source.
Your (ANSI-only) code should compile with 2.9 without any changes (or with
very few changes, please let us know if you need to change anything non
trivial and/or not mentioned in docs/changes.txt already) and the redraw
logic in wxGTK has been changed in this version. Maybe it helps with your
bug.
Thomas Spitzer
> between the traces. This really does look like a problem with wxGTK
> refresh/redrawing and latest GTK+. Unfortunately I don't know what could it
> be exactly.
node, ignoring the missing line, the connection is still not made.
In my opinion it might have to do with an unexpected mouse event. The
dotted line exists from the selected node to the mouse pointer, as long
as you move inside the window. On a mouse leave event (when mouse
pointer leaves the window) it is destroyed. This is my guess.
> One proposal I do have is for you to try 2.9.0 or the latest svn source.
> Your (ANSI-only) code should compile with 2.9 without any changes (or with
> very few changes, please let us know if you need to change anything non
> trivial and/or not mentioned in docs/changes.txt already) and the redraw
> logic in wxGTK has been changed in this version. Maybe it helps with your
> bug.
already from windows build. Maybe I'll try this, next.
But it will not help me with my main problem. Maybe I express the goal
her:
I want to provide the community with an Astade ubuntu repository for
Karmic. If wxWidgets2.9 would be the solution, my next question would be
"How to generate Ubuntu packages of 2.9" ;-)
I even thought of changing Astade that way, that it can compile with
wxWidgets2.8 unicode. But this is useless work, because from 2.9 there
is not such a difference any more.
Regards
Thomas
Thomas Spitzer
I made the two traces and they are definitely different! (Look at the
two attached pictures)
You can see, it is as expected! There is a useless MouseLeave event not
long after creating the EdgePointer. Any idea, where this is coming
from??
Regards
Thomas
Mark Gollahon
I've got quite a conundrum here. I've got four windows with "maps" of
objects in them. I can drag an object from one of the maps into another
window to see its properties. However, if a fumble-finger user starts
dragging an object to a different window but instead of making it to the
other window they drop the object in the same window as the originating
map, the map gets replaced. I'd like to keep that from happening by not
allowing the drop if the object being dragged originated from the window
it is being dropped in. Can anyone versed in wx's DnD give me a clue as
to how to solve this?
Things I've tried:
1) Overriding wxDragTarget::OnDragOver(). The problem here is that I
don't know how to get the wxDragSource or its owning window to know if
the wxDragTarget is associated with it.
2) Overriding wxDragSource::GiveFeedback(). The problem here is that I
don't know how to get the wxDragTarget the user's mouse is currently
over.
3) Recording the originating window in the DnD string and trying #1.
The problem here is that I can't seem to get the DnD string during
OnDragOver().
A pointer would be greatly appreciated!
Regards,
Mark Gollahon
Exacq Technologies, Inc.
Mark Gollahon
issuing the drag before executing the drag, then add it back once the
drag is over....
Vadim Zeitlin
MG> Can anyone versed in wx's DnD give me a clue as to how to solve this?
Use your custom data object and initialize it with information about the
source/origin region it was dragged from, then check if it's the same as
the current one in your wxDropTarget::OnEnter().
Vadim Zeitlin
TS> I made the two traces and they are definitely different! (Look at the
TS> two attached pictures)
[Please notice that sending 100KB attachments to the list is rather frowned
upon, please put them on your project web server and provide link to them
if possible]
TS> You can see, it is as expected! There is a useless MouseLeave event not
TS> long after creating the EdgePointer. Any idea, where this is coming
TS> from??
No, sorry, not really. But this should be relatively easy to debug as you
can always put a breakpoint at the location (in src/gtk/window.cpp) where
this event is generated and see where is it coming from...
Mark Gollahon
drop target from the dragged-from window while the drop operation is in
process; then restore it afterwards.
However, your solution still won't provide visual feedback to the user
that a drop cannot occur because the drop target cannot get the data
until /after/ the drop occurs. By that time the visual indicator is of
no effect. Furthermore, the drop source *could* know about the source
window, and the drop target *could* know about the target window; but
drop source doesn't know about drop target or visa-versa until the drop
occurs.
Regards,
MarkG