orbit-python-list

Re: [Orbit-python-list] orbit-python and orbit2 anywhere?

[<ml...@kn...>]

On Mon, Jan 28, 2002 at 08:58:26PM +0100, Johan Dahlin wrote:
> mån 2002-01-28 klockan 20.23 skrev Christian Robottom Reis:
> > On Mon, 28 Jan 2002, ml wrote:
> > 
> > > is there a port of orbit-python to orbit2?  AFAIK,
> > > "corbaloc:", "corbaname:" etc. are not supported by
> > > orbit1.

For those who need only "corbaloc:" (that's ok for me at the
moment) without all the other goodies of orbit2: I hacked a
few lines of python code to emulate that behaviour.  I just
split the "corbaloc:" string and call genior (from omniORB3)
using popen with the parameters from the string and read the
generated IOR back to the client program.  

Cheers!

Re: [Orbit-python-list] Do activate_object_with_id and -ORBIPv4Port really work?

[Christian R. R. <ki...@as...>]

On Thu, 31 Jan 2002, ml wrote:

> > I hacked 0.3.1 to fix this, but it could be that the parameter is not
> > being passed along. I'm not sure if that's your problem. What happens if
> > you don't pass those in?
>
> The first two arguments do have the expected effect, the
> third not.

It could be an orbit bug; can you try and bring it up with them?

We may not be passing in the parameter, but I find that very hard to
believe if the others do and we don't manipulate them. Changing order does
nothing, I assume?

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] Do activate_object_with_id and -ORBIPv4Port really work?

[<ml...@kn...>]

On Wed, Jan 30, 2002 at 08:41:09PM -0200, Christian Robottom Reis wrote:
> On Tue, 29 Jan 2002, ml wrote:
> 
> > sys.argv.append("-ORBIIOPIPv4=1")
> > sys.argv.append("-ORBIIOPUSock=0")
> > sys.argv.append("-ORBIPv4Port=7777")
> 
> I hacked 0.3.1 to fix this, but it could be that the parameter is not
> being passed along. I'm not sure if that's your problem. What happens if
> you don't pass those in?

The first two arguments do have the expected effect, the
third not.

Cheers.

Re: [Orbit-python-list] ORBit-python 1.99.0 released

[<ml...@kn...>]

Hi,

On Thu, Jan 31, 2002 at 10:57:48AM +0100, Johan Dahlin wrote:
> I just uploaded orbit-python-1.99.0.tar.gz to ftp.gnome.org (actually
> James Henstridge did).

Thanks for helping debugging by leaving a 1.2M core file
under test-suite/ :-)

Anyway, I tried compiling, but idl.c and types.c failed
because of a problem with 'sublabels'.  Has it to be
CORBA_long or CORBA_any or what?  Maybe it's because I
have still orbit2 2.3.102 and not 2.3.103?

Cheers!

Re: [Orbit-python-list] How to use IDL unions with orbit-python?

[Christian R. R. <ki...@as...>]

On Thu, 24 Jan 2002, ml wrote:

> module M {
>   struct S {
>     enum E1 { A, B } e;
>   };
>   union U switch (enum E2 { A, B }) {
>     case A: ...;
>     case B: ...;
>   };
>   ...
> };

We had a bug reported on unions not working as expected a while back.
While this may not have any impact on you, I'm not very sure. If you are
following the spec and it doesn't work, look at bug 13 (on
bugzilla.sault.org) and if that's it, add a comment. Otherwise, file a new
one and I'll get to looking at it.

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] Spec non-compliance re structures

[Christian R. R. <ki...@as...>]

Filed bug 14 on this. It helps me remember stuff.

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] Spec non-compliance re unions

[Christian R. R. <ki...@as...>]

On Mon, 29 Oct 2001, Chip Richards wrote:

> Tack requested that I send details about comments I made on IRC, so here they
> are.

Chip, if you could open an account on http://bugzilla.sault.org, I've
opened bug 13 (and would have CCed you on it).

I would like to know, has this been fixed with Johan's recent fix?

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

[Orbit-python-list] ORBit-python 1.99.0 released

[Johan D. <jd...@te...>]

Hi

I just uploaded orbit-python-1.99.0.tar.gz to ftp.gnome.org (actually
James Henstridge did).

When the mirrors is synced it'll be availible at:

ftp://ftp.gnome.org/pub/GNOME/earthquake/sources/orbit-python/orbit-python-1.99.0.tar.gz

This is a work in progress, so it's not suite for real work yet.

Only thing that differs from the stable series is that 1.99.0 is ported
ORBit2

Only 10 tests failes, but we still have huge memory leaks, anyone up to
this?

Requirements:

glib    1.3.12
linc    0.1.6
libIDL  0.7.1
ORBit2  2.3.103

And if you wish to build from CVS, check out the module orbit-python in
gnome cvs.
(Note that automake 1.5 and autoconf 2.52 is required for CVS).

---
Johan Dahlin

Re: [Orbit-python-list] Do activate_object_with_id and -ORBIPv4Port really work?

[Christian R. R. <ki...@as...>]

On Tue, 29 Jan 2002, ml wrote:

> sys.argv.append("-ORBIIOPIPv4=1")
> sys.argv.append("-ORBIIOPUSock=0")
> sys.argv.append("-ORBIPv4Port=7777")

I hacked 0.3.1 to fix this, but it could be that the parameter is not
being passed along. I'm not sure if that's your problem. What happens if
you don't pass those in?

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] non-blocking patch

[Christian R. R. <ki...@as...>]

On Wed, 30 Jan 2002, Adam Olsen wrote:

> I spoke too soon.  Attempting to use a callback deadlocks. :(
> (I'm attaching my code incase someone else has a clue.)

Have you tried not sleeping? Alas, that is the problem with mainiteration
lacking I think.

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] non-blocking patch

[Duncan G. <dg...@uk...>]

On Wednesday 30 January, Adam Olsen wrote:

> Attached is a simple patch which adds an iterate() function, which is
> like run() only it returns immedietly instead of blocking. :)

The standard name for this function is perform_work(). It has a
companion function called work_pending() that returns true if there is
work to be done. Any ORBit-Python addition to support it should use
the standard names.

Cheers,

Duncan.

-- 
 -- Duncan Grisby  \  Research Engineer  --
  -- AT&T Laboratories Cambridge          --
   -- http://www.uk.research.att.com/~dpg1 --

Re: [Orbit-python-list] non-blocking patch

[Adam O. <rh...@d2...>]

On Wed, Jan 30, 2002 at 12:01:40PM -0200, Christian Robottom Reis wrote:
> On Wed, 30 Jan 2002, Adam Olsen wrote:
> 
> > try:
> >     while 1:
> >         orb.iterate()
> > except:
> >     pass
> 
> Could you provide an example usage of this feature?

I spoke too soon.  Attempting to use a callback deadlocks. :(
(I'm attaching my code incase someone else has a clue.)


-- 
Adam Olsen, aka Rhamphoryncus

Re: [Orbit-python-list] non-blocking patch

[Christian R. R. <ki...@as...>]

On Wed, 30 Jan 2002, Adam Olsen wrote:

> try:
>     while 1:
>         orb.iterate()
> except:
>     pass

Could you provide an example usage of this feature?

I am unfamiliar with the mainloop in glib and orbit, so i dont feel
comfortable to comment. However, I do know that the signal handler there
is going to cause problems.

Also, you reported that orb.shutdown() is going to cause segfaults. It
would be nice to avoid that as well; your idea for a global check sounds
good. Tack?

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

[Orbit-python-list] non-blocking patch

[Adam O. <rh...@d2...>]

Attached is a simple patch which adds an iterate() function, which is
like run() only it returns immedietly instead of blocking. :)

It works with my very limited testing (while 1: orb.iterate()), but
like shutdown(), it doesn't bother to check that the orb is actually
running, and can cause segfaults because it is.  It'd be very nice if
it threw an exception, since that'd allow you to do:
try:
    while 1:
        orb.iterate()
except:
    pass

which'd act the same as simply calling orb.run().  except it doesn't
sleep of course :)


-- 
Adam Olsen, aka Rhamphoryncus

[Orbit-python-list] Do activate_object_with_id and -ORBIPv4Port really work?

[<ml...@kn...>]

Hi,

I want to have a CORBA client that does not need to know the
IOR of the server, but without the overhead of a CORBA
naming service.  If the server has a single factory object
with a fixed ID the server uses a fixed client-known TCP
port, the client should reach the service using sth. like:

$ ./client corbaloc::192.168.10.20:7777/Echo

Unfortunately my orbit-python based server doesn't do what I
hoped for:

#!/usr/bin/env python

import random, CORBA
from _GlobalIDL__POA import Echo
import sys

class Echo_Impl(Echo):

    def echoString(self, astring):
        outnum = random.randint(0, 100) / 100.0
        print "[server] %s -> %f" % (astring, outnum)
        return self._this(), outnum

sys.argv.append("-ORBIIOPIPv4=1")
sys.argv.append("-ORBIIOPUSock=0")
sys.argv.append("-ORBIPv4Port=7777")
orb = CORBA.ORB_init(sys.argv, CORBA.ORB_ID)
servant = Echo_Impl()
poa = orb.resolve_initial_references("RootPOA")
poa.activate_object_with_id("Echo", servant)
print orb.object_to_string(servant._this())
poa.the_POAManager.activate()
orb.run()

If I look at the ior string (using catior, ior-decode, or
ior-decode-2), the TCP port is e.g. 1188, not 7777, and the
object key is no "Echo", but I don't know what.  I believe
that I misunderstood at least two CORBA or orbit-python
concepts, right?  Could someone enlighten me, please?

Thanks in advance!

Re: [Orbit-python-list] How to use IDL unions with orbit-python?

[Duncan G. <dg...@uk...>]

On Saturday 26 January, ml wrote:

> I should have written, that I already read the spec, and
> also found some examples both for orbit-python and
> omniORBpy, but no document showed the use of "embedded"
> enum values e.g. in a union:
> 
> module M {
>   union U switch (enum E { A, B }) {
>     case A: ...;
>     case B: ...;
>   };
> };

The enum is in the scope of the union, so you should be able to do

  u = M.U(M.A, foo)

where foo is the right type for case A.

I believe ORBit-Python fails to cope with the nested enum. See my
mailing list posting:

  http://sourceforge.net/mailarchive/forum.php?thread_id=213405&forum_id=4733

> So it's more a problem with enums than with unions,
> because here I have the same problem:
> 
> module M {
>   struct S {
>     enum E { A, B } X;
>   };
> };

Here, the enum is in the scope of S, so you should be able to do

  s = M.S(M.S.B)

Again, it seems to be broken with ORBit-Python.

Cheers,

Duncan.

-- 
 -- Duncan Grisby  \  Research Engineer  --
  -- AT&T Laboratories Cambridge          --
   -- http://www.uk.research.att.com/~dpg1 --

Re: [Orbit-python-list] How to use IDL unions with orbit-python?

[Jochen V. <jv...@we...>]

Hello,

thank you for the quick fix.

On Mon, Jan 28, 2002 at 09:02:01PM +0100, Johan Dahlin wrote:
> m=E5n 2002-01-28 klockan 20.07 skrev Jochen Voss:
> > Part of the problem may be the bug, which is
> > described at
> >=20
> >     http://bugs.debian.org/123848
> >=20
> > The access to union members does not follow
> > the CORBA python language mapping specification.
> > This really should be fixed.
> >=20
> Could you try the attached patch and see if it works better for you?

Yes, I tried it and it is a big improvement.
The _d and _u attributes have the correct names, now.

But there are still problems left.
I will use the following IDL file to explain.

    module TestUnion {
      interface Frosch {
	enum E2 { A, B };
	union MyUnion switch (E2) {
	case A:
	  long x;
	case B:=20
	  string s;
	};

	attribute MyUnion U;
      };
    };

Now, Section 1.3.5 of the CORBA python language mapping
says:

    Union types are mapped to classes with two attributes.
    [ ... _d and _v are explained ... ]
    For each branch, there is an additional attribute,
    which can only be accessed if the branch has been set.

This additional attribute seems to be no implemented:

    voss@tatonka [~/src/python/test-union] python
    Python 2.1.2 (#1, Jan 18 2002, 18:05:45)=20
    [GCC 2.95.4  (Debian prerelease)] on linux2
    Type "copyright", "credits" or "license" for more information.
    >>> import CORBA, TestUnion
    >>> x=3DTestUnion.Frosch.MyUnion(TestUnion.Frosch.B,"Hallo")
    >>> x
    <__main__.MyUnion instance at 0x80a606c>
    >>> x._d
    1
    >>> x._v
    'Hallo'
    >>> x.s=20
    Traceback (most recent call last):
      File "<stdin>", line 1, in ?
    AttributeError: MyUnion instance has no attribute 's'

The way I read section 1.3.5, x.s should have been 'hallo'.
Next: the specification continues:

    [...]
    * If the discriminant is not listed [in a case label],
      and there is no default, the value is "None".

This does not work, too:

    voss@tatonka [~/src/python/test-union] python
    Python 2.1.2 (#1, Jan 18 2002, 18:05:45)=20
    [GCC 2.95.4  (Debian prerelease)] on linux2
    Type "copyright", "credits" or "license" for more information.
    >>> import CORBA, TestUnion
    >>> x=3DTestUnion.Frosch.MyUnion()
    >>> x._d=3D2
    AttributeError: MyUnion instance has no attribute '_v'
    >>> x._v
    Traceback (most recent call last):
      File "<stdin>", line 1, in ?
    AttributeError: MyUnion instance has no attribute '_v'

Even if it is simply not inplemented, the first error
message looks strange.

I hope this helps,
Jochen
--=20
                                         Omm
                                      (0)-(0)
http://www.mathematik.uni-kl.de/~wwwstoch/voss/privat.html

Re: [Orbit-python-list] How to use IDL unions with orbit-python?

[Johan D. <jd...@te...>]

m=E5n 2002-01-28 klockan 20.07 skrev Jochen Voss:
> Hello,
>=20
> On Thu, Jan 24, 2002 at 12:20:45AM +0100, ml wrote:
> > Hi,
> >=20
> > [...]
> >=20
> > How can I access A and B in client and server code in
> > both cases?  I tried different variants like M.A and
> > M.U.A, but didn't succeed.
>=20
> Part of the problem may be the bug, which is
> described at
>=20
>     http://bugs.debian.org/123848
>=20
> The access to union members does not follow
> the CORBA python language mapping specification.
> This really should be fixed.
>=20
Could you try the attached patch and see if it works better for you?

> Jochen
> --=20
>                                          Omm
>                                       (0)-(0)
> http://www.mathematik.uni-kl.de/~wwwstoch/voss/privat.html

Re: [Orbit-python-list] orbit-python and orbit2 anywhere?

[Johan D. <zil...@ho...>]

m=E5n 2002-01-28 klockan 20.23 skrev Christian Robottom Reis:
> On Mon, 28 Jan 2002, ml wrote:
>=20
> > is there a port of orbit-python to orbit2?  AFAIK,
> > "corbaloc:", "corbaname:" etc. are not supported by
> > orbit1.
>=20
> There is, check out gnome CVS, module "orbit-python" IIRC. Am I right,
> Johan?
Yepp.

export CVSROOT=3D':pserver:ano...@an...:/cvs/gnome'
cvs login
cvs -z3 co orbit-python

Note. The module has the same name as the one in
orbit-python.sourceforge.net

>=20
> Take care,
> --
> Christian Reis, Senior Engineer, Async Open Source, Brazil.
> http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL
>=20

Re: [Orbit-python-list] orbit-python and orbit2 anywhere?

[Christian R. R. <ki...@as...>]

On Mon, 28 Jan 2002, ml wrote:

> is there a port of orbit-python to orbit2?  AFAIK,
> "corbaloc:", "corbaname:" etc. are not supported by
> orbit1.

There is, check out gnome CVS, module "orbit-python" IIRC. Am I right,
Johan?

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] How to use IDL unions with orbit-python?

[Jochen V. <jv...@we...>]

Hello,

On Thu, Jan 24, 2002 at 12:20:45AM +0100, ml wrote:
> Hi,
>=20
> [...]
>=20
> How can I access A and B in client and server code in
> both cases?  I tried different variants like M.A and
> M.U.A, but didn't succeed.

Part of the problem may be the bug, which is
described at

    http://bugs.debian.org/123848

The access to union members does not follow
the CORBA python language mapping specification.
This really should be fixed.

Jochen
--=20
                                         Omm
                                      (0)-(0)
http://www.mathematik.uni-kl.de/~wwwstoch/voss/privat.html

[Orbit-python-list] orbit-python and orbit2 anywhere?

[<ml...@kn...>]

Hi,

is there a port of orbit-python to orbit2?  AFAIK,
"corbaloc:", "corbaname:" etc. are not supported by
orbit1.

Thanks in advance!

Re: [Orbit-python-list] Any activity?

[Christian R. R. <ki...@as...>]

On Wed, 23 Jan 2002, Roland Mas wrote:

>   Is anyone else but me interested in fixing these bugs?  Come on
> guys, let's wake up and make ORBit-Python what it deserves to be!

I'm awake, just very very busy. Promise to shape up in the following
weeks.

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] Any activity?

[Christian R. R. <ki...@as...>]

On Fri, 25 Jan 2002, Roland Mas wrote:

> Roland Mas (2002-01-25 11:21:57 +0100) :
>
> Well, it's done.  Know that this fix shows a memory leak, though.  It
> doesn't exactly /introduce/ this leak, since the conditions leading to
> it previously led to a segfault, but there it is.  I had a fix for
> that leak, but it was a fix for ORBit, and I was told by one ORBit
> guru that it was wrong.  Apparently it's okay for applications
> (including ORBit-Python) to fiddle with the deep innards of ORBit
> proper (including ...->_private members of structs).  I'm looking for
> another, ORBit-Python-only, fix.

Have you tested with ORBit-0.5.12? At least from what I remember, we had a
lot of trouble because somebody was murdering ->_private members of
deactivated servants.

I can't see a way to fix that without having the ORBit folks reversing
that patch.

Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 272 3330 | NMFL

Re: [Orbit-python-list] How to use IDL unions with orbit-python?

[<ml...@kn...>]

Hallo,

thanks for the quick reply!

On Fri, Jan 25, 2002 at 06:02:48PM +0100, Roland Mas wrote:
> I'm afraid I can't give you a definitive answer since I don't use
> unions, but you might consider reading the CORBA Python language
> mapping specification.  I seem to remember it explains precisely the

I should have written, that I already read the spec, and
also found some examples both for orbit-python and
omniORBpy, but no document showed the use of "embedded"
enum values e.g. in a union:

module M {
  union U switch (enum E { A, B }) {
    case A: ...;
    case B: ...;
  };
};

So it's more a problem with enums than with unions,
because here I have the same problem:

module M {
  struct S {
    enum E { A, B } X;
  };
};

I think I have to put the enum outside of the union
or the struct, but that's not so nice.

Cheers!