[Parti-discuss] mac osx support (and probably others)

[Nathaniel Smith]

On Fri, Jan 8, 2010 at 11:42 PM, Antoine Martin <antoine at nagafix.co.uk> wrote:
> I was doing some testing on Mac OSX and I thought I would see if xpra ran on
> it.
>
> It was surprisingly easy to get it going.
> The client "just works"! (if you can be bothered installing all the required
> dependencies via macports)

Cool, good to know.

> For the server, the only change required is to replace "/proc/self/fd" with
> "/dev/fd" in xpra/scripts/server.py
>
> Maybe we should be using "/dev/fd" as a more portable way to get to the fds?
> This works on all the Linux distros I have access to and on Mac OSX (will
> try Solaris / FreeBSD next time I am on one of those test systems)

Committed.

> Now, admittedly that does not really give you a functional xpra...
> It works for xlogo, but xcalc looks all mangled up...
> And the keyboard does not work at all (xterm looks ok though):
>
> Unhandled error while processing packet from peer
> Traceback (most recent call last):
>  File
> "/opt/local/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/xpra/protocol.py",
> line 143, in _consume_packet
>    self._process_packet_cb(self, decoded)
>  File
> "/opt/local/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/xpra/server.py",
> line 712, in process_packet
>    self._packet_handlers[packet_type](self, proto, packet)
>  File
> "/opt/local/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/xpra/server.py",
> line 649, in _process_key_action
>    self._keycode(keyname), depressed)
>  File
> "/opt/local/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/xpra/server.py",
> line 434, in _keycode
>    return self._keymap.get_entries_for_keyval(keyval)[0][0]
> TypeError: 'NoneType' object is unsubscriptable

I'm not sure what's going on with xcalc, but this error looks like
another instance where the X server is getting started without a
useful keymap, and our keyboard code is still too stupid for that.
Loading a keymap by hand might be a workaround for now...

> PS: Nathaniel, I will take a look at your win32 socket code as soon as I can

Thanks!

-- Nathaniel