Created attachment 160801 [details]
This debug.out possible show why happening this problem

From a quick glance, I'm not seeing anything in your debug.out illustrating why the problem is happening.

My guess is that it might be a speech dispatcher bug. Jon, is this something you can look at?

Lowering the status to critical based on the GNOME official definitions:

Blocker  	Blocks development and/or testing work
Critical 	Crashes, causes loss of data, or is a severe memory leak

Hi,

I had a try at reproducing this report on lucid and karmic, but no success.

I did the following:

1. made sure the checkbox is checked in the speech tab.
2. opened gedit, and when to the first item in the file menu.
3. performed a where_am_I (on the 'new' menu item)
4. arrowed down to open, and did where am I there.
5. exit the menu, 
6. opened orca preferences.
7. changed the checkbox to unchecked.
8. repeated 2-4, without any problems.
9. Restored the checkbox to checked (this is the default)

Is your system fully updated?

Did you upgrade from karmic to lucid, or did you install it fresh?

No one has replied on the mailing list yet, saying that they can confirm this.

Thanks.

-Jon

Another question running through my head in light of Jon's findings is: Does this happen with English too, or just Hungarian?

Hy Jon and Joanie,

Jon, this is very interesting. Of course my system is full updated. I doed a clean Lucid installation, not Karmic upgrade to Lucid.
When you try reproduce the problem, you use Speech-dispatcher and Espeak speech sinthesizer driver?
My Orca speech settings are following:
Speech system: Speech-dispatcher:
Speech-sinthesizer: Espeak
Person: hungarian

Oldest time I doed an another this check box related problem, possible you verify this your system? When this check box is checked, and press Orca+q key combination, the Orca dialog are freezed, impossible navigate to quit dialog controls. This is true if the speech system is Speech-dispatcher, speech sinthesizer driver is Espeak. If this check box is unchecked, this problem is not happening my machine, similar with where am I command problem. Here this oldest bugreport link, the where am I problem simptoms is very similar with this bug:
https://bugzilla.gnome.org/show_bug.cgi?id=601258

I very hope this problem is only a speech related configuration issue and not real problem in Orca code. Joanie, just a moment, I look another Orca Espeak person character to possible reproduce the problem or not.

Attila

Joanmarie, unfortunately the problem is happening my machine with language undependent.
I choosed english language locale, and tryed following reproducation sptes:
1. I launched Gedit application.
2. Wroted apple text.
3. Press where am I key (Orca+Return key combination with laptop keyboard layout). When Orca spokened the actual edit box line informations, after this task is done, Orca is not talking, only help when I run orca --replace command to restart Orca.
I tryed changing speech related settings with following, but not help:
Speech-system: Speech-dispatcher (not changed)
Speech sinthesizer: Default (chanded from Espeak to first value in combo box).
Person: en (Espeak en voice)
So, this is possible not a missconfiguration issue my machine, but I not have ydea why happening this problem.

Have another ydea?

Attila

I can confirm this bug happens in Vinux 3.0 as well, which is based on Lucid, and ships Orca 2.30.1.  I have to restart orca and speech-dispatcher to get voice back.

Thanks Bill. Now we know it's not Attila's imagination. (Just kidding, Attila!)

Bill, any ideas as to where the problem may be and how to fix it?

Also please, 32/64 bit, single or multiple core?

Jon, My notebook have an Intel Dualcore T2300 1.73 GHZ processor, and I using now Ubuntu Lucid 64 bit release.
See the proc/cpuinfo data, the Cpu MHZ now reduced with 800 MHz:
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Pentium(R) Dual  CPU  T2370  @ 1.73GHz
stepping	: 13
cpu MHz		: 800.000
cache size	: 1024 KB
physical id	: 0
siblings	: 2
core id		: 0
cpu cores	: 2
apicid		: 0
initial apicid	: 0
fpu		: yes
fpu_exception	: yes
cpuid level	: 10
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good aperfmperf pni dtes64 monitor ds_cpl est tm2 ssse3 cx16 xtpr pdcm lahf_lm
bogomips	: 3458.53
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 1
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Pentium(R) Dual  CPU  T2370  @ 1.73GHz
stepping	: 13
cpu MHz		: 1733.000
cache size	: 1024 KB
physical id	: 0
siblings	: 2
core id		: 1
cpu cores	: 2
apicid		: 1
initial apicid	: 1
fpu		: yes
fpu_exception	: yes
cpuid level	: 10
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good aperfmperf pni dtes64 monitor ds_cpl est tm2 ssse3 cx16 xtpr pdcm lahf_lm
bogomips	: 3458.38
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

My desktop machine have an Athlon64 X2 3800+ processor, next comment I sending important /proc/cpuinfo data if I tested this problem my wife 32 bit Ubuntu Lucid version.

Attila

(In reply to comment #9)
> Also please, 32/64 bit, single or multiple core?

I reproduced this bug on a somewhat hacked version of Ubuntu Lucid, upgraded to kernel 2.6.32-21-generic.  I have a 4-core i7 x64 installation on a Dell Studio 1557 laptop, with 4 gig memory and a proprietary ATI graphics driver.

I did an apt-get dist-upgrade, which also incremented my kernel to linuz-2.6.32-22.  I can no longer reproduce the bug, either on my multi-core x64 installation of Vinux, based on Ubuntu Lucid, or on my other single-cpu x32 and x64 Vinux installs which I run in VirtualBox virtual machines.

This bug may have gone away somehow.

Jon, I tested this problem my wife 32 bit Ubuntu lucid system with an Athlon64 X2 Dualcore processor (3800+ MHZ, the real speed is 1800 MHZ).
My wife Lucid system is full uptodated with apt-get dist-upgrade command before the test. The used kernel is 2.6.32-22-generic both my desktop and my notebook machines.
Unfortunately the problem is reproducable both 32 bit and 64 bit uptodated Ubuntu Lucid systems my home machines.
See my desktop machine cpuinfo values:
processor	: 0
vendor_id	: AuthenticAMD
cpu family	: 15
model		: 75
model name	: AMD Athlon(tm) 64 X2 Dual Core Processor 3800+
stepping	: 2
cpu MHz		: 1000.000
cache size	: 512 KB
physical id	: 0
siblings	: 2
core id		: 0
cpu cores	: 2
apicid		: 0
initial apicid	: 0
fdiv_bug	: no
hlt_bug		: no
f00f_bug	: no
coma_bug	: no
fpu		: yes
fpu_exception	: yes
cpuid level	: 1
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt rdtscp lm 3dnowext 3dnow extd_apicid pni cx16 lahf_lm cmp_legacy svm extapic cr8_legacy
bogomips	: 1999.82
clflush size	: 64
cache_alignment	: 64
address sizes	: 40 bits physical, 48 bits virtual
power management: ts fid vid ttp tm stc

processor	: 1
vendor_id	: AuthenticAMD
cpu family	: 15
model		: 75
model name	: AMD Athlon(tm) 64 X2 Dual Core Processor 3800+
stepping	: 2
cpu MHz		: 1000.000
cache size	: 512 KB
physical id	: 0
siblings	: 2
core id		: 1
cpu cores	: 2
apicid		: 1
initial apicid	: 1
fdiv_bug	: no
hlt_bug		: no
f00f_bug	: no
coma_bug	: no
fpu		: yes
fpu_exception	: yes
cpuid level	: 1
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt rdtscp lm 3dnowext 3dnow extd_apicid pni cx16 lahf_lm cmp_legacy svm extapic cr8_legacy
bogomips	: 1999.82
clflush size	: 64
cache_alignment	: 64
address sizes	: 40 bits physical, 48 bits virtual
power management: ts fid vid ttp tm stc

Attila

Jon, possible useful my following detected informations when I tested this problem my wife system, the results is equals my notebook when I check this interesting informations:
If I launch for example Gedit application, open a menu and try where am I command doing, you are absolute right, the bug is not reproducable with Gedit menu systems. So, where am I output menu formatting is not freeze Orca when the break speech into chunks between pauses checkbox is checked if a menu is opened. But:
If the caret have an edit box line with containing any number of characters, and try  do where am I operation, the problem is reproducable my both two test machine. If the caret positioned with an empty line, and do where am I operation, Orca is not freezed.

Possible have a not Speech-dispatcher compatible string with src/orca/formatting.py file with edit box related formatting if the prewious wroted useExperimentalSpeechProsody setting is true?

All, I can reproduce this bug partially, independent of the value of the check break speech into chunks.
Try the following steps:

1. Launch gedit with an empty file.

2. When the editing area is displayed, presseione the enter key to include an empty line in the file.

3. Press alt+home to go to the top of the file.

4. Perform an where I am operation.
Orca should speak text.

5. Repeat the step 4.
Now nothing is spoken.

6. Try to repeat the steps 4 and 5 many times.
The result is the same, orca speaks text and stays mute.
Sometimes orca speaks something like "cancel self speak".

7. Observe the number of lines in the file opened in the step 1.
There should be only one line in the file, but there are several.

I'm running  on a dell dual core with a 64-bit ubuntu 10.4 installed from fresh.

Let me add one more step to the steps of the previous comment.

8. Try to access the preferences page of the orca.
it is not shown and if I try to access again the orca tells me that there is already a dialog running.

Hello Jose,

Unfortunately your wroted steps can not freezed my Orca if the check box is unchecked. Possible this is not happening because I use now Orca Git Master version.
If the check box is unchecked, where am I always working my system. You use Speech-dispatcher and Espeak speech sinthesizer or another speech sinth?
Anybody possible reproduce Jose wroted steps this problem? My system is full uptodated.

Tomorrow I try your steps with an Ubuntu Lucid live CD.

Attila

Jose, you try attach a debug.out file if you reproduced your wroted steps the problem? I would like compare my first attached debug.out file result if the freeze is happening.

Attila

I can confirm everything Jose said on my system.  I have a new install of Vinux 3.0 RC4, which is an up-to-date version of Ubuntu Lucid modified to be a bit friendlier to the blind, and using Orca 2.30.1.  It's kind of freaky how the preferences dialog box is hidden.

Bill

(In reply to comment #17)
> Jose, you try attach a debug.out file if you reproduced your wroted steps the
> problem? I would like compare my first attached debug.out file result if the
> freeze is happening.
> 
> Attila

(In reply to comment #16)
> Hello Jose,
> 
> Unfortunately your wroted steps can not freezed my Orca if the check box is
> unchecked. Possible this is not happening because I use now Orca Git Master
> version.
> If the check box is unchecked, where am I always working my system. You use
> Speech-dispatcher and Espeak speech sinthesizer or another speech sinth?
> Anybody possible reproduce Jose wroted steps this problem? My system is full
> uptodated.
> 

I use speech-dispatcher and I tested with espeak and ibm viavoice.
The same result with both synthesizers.

Created attachment 161098 [details]
debug output from orca 

debug output from orca that contains the result of steps suggested in comment #14

Created attachment 161099 [details]
errors while executing steps suggested in comment #14

While executing steps suggested in comment #14, I found some messages from orca.
I am not sure if they have to do with the bug.

Thank you Jose the last attachment!
I see I think very similar trace back message if this check box is checked and I try quitting Orca, I copyed I seed trace back message with another bugreport last comment, I wroted the quit dialog related bugreport link with upper comment.
I looked what type controls are producing this problem if this check box is checked if I try do where am I operation:
1. In gnome-terminal main window, the problem always present and reproducable with actual command line (prompt line). I think this is the piatspi.role_frame pause formatting output, but I am not full sure.
2. Text control for example Gedit if have any number of characters with a line. I think this is piatspy.role_text pause format output. This is true or another text control containing application (for example when I writing this comment in Firefox and do where am I output, the problem is happened), but not true for example Openoffice.org Writer text control. In Openoffice.org Writer, I can not reproduce this problem.

Attila

Possible little bad news, not help the problem fix, but very interesting test result:
I try what happening if I install libgnome-speech7 package and switch back temporary with Gnome-Speech speech sinthesizer my Lucid system.
I try reproduce the where am I problem if this check box is checked after I choosed Gnome-Speech Speech system and Espeak Gnome-Speech driver. The where am I problem is not happening in Gnome-Speech speech backend. My another break speech into chunks between pauses/Orca quit dialog problem is not happening if I use Gnome-Speech speech sinthesizer, Orca quit dialog works perfect. But Interesting. Why do this Jose copyed invalid command freeze problem with Speech-dispatcher and why not do Gnome-Speech?

Attila

I found the problem.

it is nothing to do with orca, but with sd.
I dont know what happens to it but when it sees a new line followed by a full stop something seems to go wrong internally, or maybe we are entering a special mode.


trying this from command line:

--start--
[ch@Ms:3:~]$: ipython
Python 2.6.4 (r264:75706, Dec  7 2009, 18:43:55)
Type "copyright", "credits" or "license" for more information.

IPython 0.10 -- An enhanced Interactive Python.
?         -> Introduction and overview of IPython's features.
%quickref -> Quick reference.
help      -> Python's own help system.
object?   -> Details about 'object'. ?object also works, ?? prints more.

In [1]: import speechd

In [2]: ss = speechd.SSIPClient ('test')

In [3]: ss.speak('hello')
Out[3]: (225, 'OK MESSAGE QUEUED', ('7811',))
# i hear hello as expected
In [4]: ss.stop()

In [5]: ss.speak("hello\n.")
Out[5]: (225, 'OK MESSAGE QUEUED', ('7812',))
# i hear hell, and it stops
In [6]: ss.stop()
---------------------------------------------------------------------------
SSIPCommandError                          Traceback (most recent call last)

/home/cs3mmh/<ipython console> in <module>()

/usr/lib/python2.6/dist-packages/speechd/client.pyc in stop(self, scope)
    570
    571         """
--> 572         self._conn.send_command('STOP', scope)
    573
    574     def pause(self, scope=Scope.SELF):

/usr/lib/python2.6/dist-packages/speechd/client.pyc in send_command(self, command,
 *args)
    263         code, msg, data = self._recv_response()
    264         if code/100 != 2:
--> 265             raise SSIPCommandError(code, msg, cmd)
    266         return code, msg, data
    267

SSIPCommandError: 500: ERR INVALID COMMAND

In [7]:

--end--

we could filter for full stops after new lines, but that would just work around the issue rather than fix it. Trev said he was willing to look into it from opentts point of view.

-Jon

Jon, you are the best. :-):-)
Look for example the Orca quit dialog question message, now I think I understand why happen the quit dialog problem if the check box is checked if Speech-dispatcher not like the too lot of \n simbols:
msgid ""
"<b><big>Quit Orca?</big></b>\n"
"\n"
"This will stop all speech output and screen magnification.\n"
"\n"
I copyed this original english language message with orca/po/hu.po file.

Jon, have any ydea how can possible fix this problem in Lucid if this is not an Orca related problem? I don't no if this is a Speech-dispatcher related problem, developers do Speech-dispatcher update in Lucid or not. I think Opentts is not part of Lucid, or I not see full clean what the difference with two project.

Attila

(In reply to comment #25)
> Jon, you are the best. :-):-)

Not at all, i could not have done it without you and Jose

A team is always better :)

> Look for example the Orca quit dialog question message, now I think I
> understand why happen the quit dialog problem if the check box is checked if
> Speech-dispatcher not like the too lot of \n simbols:
> msgid ""
> "<b><big>Quit Orca?</big></b>\n"
> "\n"
> "This will stop all speech output and screen magnification.\n"
> "\n"
> I copyed this original english language message with orca/po/hu.po file.
> 
> Jon, have any ydea how can possible fix this problem in Lucid if this is not an
> Orca related problem? I don't no if this is a Speech-dispatcher related
> problem.

Yes, we can work around it from orca, until it gets fixed in opentts/sd.

Patch coming soon.

-Jon

Created attachment 161127 [details] [review]
rev 1

This patch should get rid of the traceback that we saw.
Attila could you please check this, and also check if it also fixes the quit problem for you?

Thank you.

-Jon

ok, the short story, this is a bug in the python module that talks to speech dispatcher. The longer story is that in the ssip protocal sending a line containing only '.' means you are done sending a chunk of text.  If you would like to send such a line you should escape it with a second dot, and if you wanted to send the line '..' you'd send '...' and so on.  However the python module should deal with this for you.  I believ that this has ben fixed in opntts, I'll take a look and tell you guys what I find.

(In reply to comment #27)
> Created an attachment (id=161127) [details] [review]
> rev 1
> 
> This patch should get rid of the traceback that we saw.

Yes, works fine with a small side effect.
When I execute my steps suggested in comment #14, I hear also a dot.
something like: text dot.
It is not a big problem for me.

Ok, it looks like this is fixed in opentts master at git://opentts.org/opentts.git, to test this, you'll need to patch orca to import opentts, the patch is in the opentts mailing list, I'll send details to the list.   As another check, can I see the code for the send data function in the speechd module you have?

Thanks!

(In reply to comment #29)
> Yes, works fine with a small side effect.
> When I execute my steps suggested in comment #14, I hear also a dot.
> something like: text dot.

I believe thats because your punctuation level is set to announce dots?
The full stop was added as part of bgo#575417
hmm, thinking aloud here, would bgo#575417 still be ok if we didnt add any full stops after newlines. ok, another patch coming up.
Need Attila's feedback on this one to make sure that we havent effected bgo#575417.

Thanks everyone.

(In reply to comment #30)
Thanks Trev for looking into this.
The code we have lives in src/orca/speechdispatcherfactory.py

if you follow the flow from def speak(), you can see exactly what is currently being done.

Created attachment 161144 [details] [review]
Rev 2

Please test.

Thanks.

(In reply to comment #32)
> Created an attachment (id=161144) [details] [review]
> Rev 2
> 
> Please test.
> 
> Thanks.
Works fine without problems.
You deserve a beer!
Thanks!

Jon, I tested the second patch, works very good my system.
My this check box related problems are solwed if the break speech into chunks between pauses is checked. Where am I works correct again with Gedit and gnome-terminal, and Orca quit dialog is works perfect.
Now, I use Orca again with this checkbox checked setting, thank you very mutch the fix, because this checked setting is a better speech output intonation with important informations with Espeak, I very like oldest time this setting.

Thank you everybody the help fixing this bug.

Attila

Thanks All!

Joanie, should this go into gnome-2-30?

-Jon

commited to master:
http://git.gnome.org/browse/orca/commit/?id=68055358d0b28c72bcf638d9897709969a88e8c9

*** Bug 601258 has been marked as a duplicate of this bug. ***

Thank you Jon the master commit.
I very hope possible to commit your patch with 2.30 branch, but if Joanie not agree this, I agree. :-):-)
Prewious after I see your commit I changed the bug 601258 Orca quit dialog related bug status to resolved fixed status, because this bug are not present now and resolved your committed patch.

Thank you every body,

Attila

(In reply to comment #35)
> Joanie, should this go into gnome-2-30?

Given the severity of the problem and the fact that the patch looks isolated and safe, unless you can think of some reason not to Jon I'd say yes please and thank you. :-)

Thanks for this patch!  And, thanks to everyone for tracking the bug down.  Jon and Joanmarie, if Orca can release a Orca 2.30.2 that includes this patch, I'll upload it to the Vinux/Lucid Testing PPA, and get some users testing it right away, and assuming it goes well for a week or two, we can put it in the main Vinux/Lucid repository.

Thanks,
Bill

The next stable release of Orca (2.30.2) is June 23rd. We follow the GNOME release schedule. [1] If Jon sees no reason why his fix should not be put back into the gnome-2-30 branch, it will be included in the 2.30.2 release.

[1] http://live.gnome.org/TwoPointThirtyone#Schedule

ok, I've opened http://opentts.org/issue/23 to deal with the bug in the opentts module, our eventual plan I believe is to move to pyrex, and then we can just worry about the C which I understand lol, but maybe somebody else can help with the python.

Trev

Hey,

Loks like I never noted it here, but the bug that causes this problem is fixed in opentts git.

Trev