From mitch at ncsa.uiuc.edu  Fri Nov  7 15:33:50 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage
Message-ID: <3.0.5.32.20031107153350.03433788@pop.ncsa.uiuc.edu>

Hi, folks -- I have a question about the Perl debugger, which I've
essentially never actually used.

How do I go about using it to find out where CPU usage is occurring?  I've
got a very large, very complex script that's soaking up *way* too much CPU
(but not leaking memory -- already checked for that) for what it's doing,
but I can't figure out where/how it's happening.

Any ideas on how to get the debugger to show me where all the CPU is going?

Mitch
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From brandon at ojctech.com  Fri Nov  7 16:28:27 2003
From: brandon at ojctech.com (Brandon Bowersox)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage
Message-ID: <5.2.0.9.0.20031107162759.02ccbc20@localhost>

Mitch,

The Perl Profiler is the tool for you and it is distributed with Perl.  See
http://search.cpan.org/~nwclark/perl-5.8.2/ext/Devel/DProf/DProf.pm
http://search.cpan.org/~nwclark/perl-5.8.2/utils/dprofpp.PL

Basically, to create a profile file, run your code like:
perl -d:DProf script.pl (on the command line)
OR
#!/usr/bin/perl -d:DProf (at the top of your script)

Then the dprofpp command will give you a report.  My favorite reports are:
dprofpp tmon.out (the basic report including CPU time)
dprofpp -S tmon.out (merged subroutine tree)

Brandon

--
Brandon Bowersox
brandon@ojctech.com
OJC Technologies, Inc.
www.ojctech.com, 217-278-3933


At 03:33 PM 11/7/2003 -0600, Mitch Kutzko wrote:
>Hi, folks -- I have a question about the Perl debugger, which I've
>essentially never actually used.
>
>How do I go about using it to find out where CPU usage is occurring?  I've
>got a very large, very complex script that's soaking up *way* too much CPU
>(but not leaking memory -- already checked for that) for what it's doing,
>but I can't figure out where/how it's happening.
>
>Any ideas on how to get the debugger to show me where all the CPU is going?
>
>Mitch
>--
>Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
>http://hobbes.ncsa.uiuc.edu/
>_______________________________________________
>Champaign-Urbana mailing list
>Champaign-Urbana@mail.pm.org
>http://mail.pm.org/mailman/listinfo/champaign-urbana


From mitch at ncsa.uiuc.edu  Fri Nov  7 16:44:52 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage
Message-ID: <3.0.5.32.20031107164452.035c5d60@pop.ncsa.uiuc.edu>

Wow -- What quick, excellent responses!  Thank you, all!

Ok -- Have found and played with the profiler and gotten results.  I was
mildly concerned that it wouldn't work properly with this script, as it's
actually one that doesn't have an ending -- It's designed to run
open-endedly for literally months on end.  Had problems with profiling via:

	$ perl -d:DProf beacon.pl

and then trying to run the profiler via:

	$ /usr/bin/dprofpp -u

That gave back:

	Garbled profile, unexpected exit time stamp at /usr/bin/dprofpp line 704,
<fh> line 680.


However, was able to run things just fine via:

	$ /usr/bin/dprofpp -u -p beacon.pl

and got the output below back.   

Now, can anyone tell me if there's an obvious way to tell where wasted
(idle?  No obvious -sleep- traps in the code, but you never know...)  CPU
is going?  

I can see that I spend most of my time in Net::RTP::rtp_recv, but that's
correct, since this is an RTP client, and spends almost all of its time
talking to the other RTP clients -- It just shouldn't be soaking up 60-80%
of my CPU doing it, that's all.

This is literally the same library that underlies vic and rat, if anybody
here has done any videoconferencing stuff...  If you've used either of
those, you know they can send and receive *LOTS* more data than this
without even being particularly noticeable.

Thanks again!

Mitch


----------------------

Total Elapsed Time = 257.3488 Seconds
         User Time = 124.1104 Seconds
Exclusive Times
%Time ExclSec CumulS #Calls sec/call Csec/c  Name
 82.4   102.2 118.95   5489   0.0186 0.0217  Net::RTP::rtp_recv
 7.73   9.589  9.582   5318   0.0018 0.0018  Net::RTP::rtp_free_rtp_packet
 3.55   4.401 17.075  49029   0.0001 0.0003  main::rtp_event_handler
 2.60   3.228  3.514    131   0.0246 0.0268  Net::RTP::rtp_send_ctrl
 1.80   2.229  2.309   5358   0.0004 0.0004  Net::RTP::rtp_update
 0.39   0.480  1.548   5396   0.0001 0.0003  IO::Socket::new
 0.31   0.390  0.331  39145   0.0000 0.0000  Net::RTP::rtp_get_rr
 0.28   0.350  1.882   5396   0.0001 0.0003  IO::Socket::INET::new
 0.27   0.330  0.256  49029   0.0000 0.0000  Net::RTP::rtp_event_type_get
 0.26   0.320  0.320      6   0.0533 0.0533  main::output_data_as_web_page
 0.25   0.310  0.302   5396   0.0001 0.0001  Symbol::gensym
 0.23   0.290  2.156   5391   0.0001 0.0004  IO::Socket::accept
 0.23   0.280  2.408   5318   0.0001 0.0005  main::receive_tcp_reports
 0.21   0.260  0.563   5396   0.0000 0.0001  IO::Handle::autoflush
 0.19   0.240  0.276   5396   0.0000 0.0001  SelectSaver::new
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From mitch at ncsa.uiuc.edu  Fri Nov  7 19:44:46 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Re: Perl Debugger and finding CPU Usage
Message-ID: <3.0.5.32.20031107194446.00fb8b38@pop.ncsa.uiuc.edu>

Hi again -- Ok, here's output collected for a bit more than 36 minutes,
using the -r and -I options.

Does anything here leap right out and grab anybody as particularly
meaningful?  It appears this is just telling me where the script spends its
time, and not so much about how much CPU is being used in various places.
I *know* where it spends its time -- I just don't know why it's using so
much CPU.

Thoughts?

Mitch


--------------------------------

Total Elapsed Time = 2202.757 Seconds
         User Time = 2202.757 Seconds
Inclusive Times
%Time ExclSec CumulS #Calls sec/call Csec/c  Name
 63.4   0.510 1398.5      1   0.5100 1398.5  main::rxtx_loop
 60.7   1171. 1337.9  44955   0.0261 0.0298  Net::RTP::rtp_recv
 7.63   41.93 168.06 445634   0.0001 0.0004  main::rtp_event_handler
 4.08   89.96 89.960  44105   0.0020 0.0020  Net::RTP::rtp_free_rtp_packet
 1.66   35.30 36.500    770   0.0458 0.0474  Net::RTP::rtp_send_ctrl
 1.00   3.750 21.950  44105   0.0001 0.0005  main::receive_tcp_reports
 0.90   19.53 19.830  44185   0.0004 0.0004  Net::RTP::rtp_update
 0.81   2.200 17.770  44719   0.0000 0.0004  IO::Socket::accept
 0.71   2.870 15.610  44734   0.0001 0.0003  IO::Socket::INET::new
 0.58   3.330 12.740  44734   0.0001 0.0003  IO::Socket::new
 0.23   1.910  5.060  44734   0.0000 0.0001  IO::Handle::autoflush
 0.17   1.300  3.670  44734   0.0000 0.0001  IO::Handle::new
 0.16   3.420  3.420 445634   0.0000 0.0000  Net::RTP::rtp_event_type_get
 0.13   2.860  2.860 363428   0.0000 0.0000  Net::RTP::rtp_get_rr
 0.12   2.340  2.660  44734   0.0001 0.0001  SelectSaver::new

--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From bhalla at uiuc.edu  Fri Nov  7 21:35:43 2003
From: bhalla at uiuc.edu (Arun Bhalla)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Re: Perl Debugger and finding CPU Usage 
In-Reply-To: <3.0.5.32.20031107194446.00fb8b38@pop.ncsa.uiuc.edu> 
Message-ID: <200311080335.hA83Zh15003385@a-night-at-the-opera.cu.groogroo.com>

Well, I think now it's time to look at the code.  I've never seen Net::RTP
before.  It looks like rtp_recv is either inefficient or is being
used incorrectly.  Perhap you can look through the Net::RTP code and
figure out what's going on.

You could use a line-level profiler like Devel::SmallProf to see which lines
are getting called the most or take the longest.  This may or may not
help you.  One major note is that SmallProf will slow down the execution
of your program, so you can expect the runtime to be much longer than usual.

Arun



Mitch Kutzko writes:
> Hi again -- Ok, here's output collected for a bit more than 36 minutes,
> using the -r and -I options.
> 
> Does anything here leap right out and grab anybody as particularly
> meaningful?  It appears this is just telling me where the script spends its
> time, and not so much about how much CPU is being used in various places.
> I *know* where it spends its time -- I just don't know why it's using so
> much CPU.
> 
> Thoughts?
> 
> Mitch
> 
> 
> --------------------------------
> 
> Total Elapsed Time = 2202.757 Seconds
>          User Time = 2202.757 Seconds
> Inclusive Times
> %Time ExclSec CumulS #Calls sec/call Csec/c  Name
>  63.4   0.510 1398.5      1   0.5100 1398.5  main::rxtx_loop
>  60.7   1171. 1337.9  44955   0.0261 0.0298  Net::RTP::rtp_recv
>  7.63   41.93 168.06 445634   0.0001 0.0004  main::rtp_event_handler
>  4.08   89.96 89.960  44105   0.0020 0.0020  Net::RTP::rtp_free_rtp_packet
>  1.66   35.30 36.500    770   0.0458 0.0474  Net::RTP::rtp_send_ctrl
>  1.00   3.750 21.950  44105   0.0001 0.0005  main::receive_tcp_reports
>  0.90   19.53 19.830  44185   0.0004 0.0004  Net::RTP::rtp_update
>  0.81   2.200 17.770  44719   0.0000 0.0004  IO::Socket::accept
>  0.71   2.870 15.610  44734   0.0001 0.0003  IO::Socket::INET::new
>  0.58   3.330 12.740  44734   0.0001 0.0003  IO::Socket::new
>  0.23   1.910  5.060  44734   0.0000 0.0001  IO::Handle::autoflush
>  0.17   1.300  3.670  44734   0.0000 0.0001  IO::Handle::new
>  0.16   3.420  3.420 445634   0.0000 0.0000  Net::RTP::rtp_event_type_get
>  0.13   2.860  2.860 363428   0.0000 0.0000  Net::RTP::rtp_get_rr
>  0.12   2.340  2.660  44734   0.0001 0.0001  SelectSaver::new
> 
> --
> Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
> http://hobbes.ncsa.uiuc.edu/
> _______________________________________________
> Champaign-Urbana mailing list
> Champaign-Urbana@mail.pm.org
> http://mail.pm.org/mailman/listinfo/champaign-urbana
> 

--
Arun Bhalla

From mitch at ncsa.uiuc.edu  Fri Nov  7 22:24:30 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Re: Perl Debugger and finding CPU Usage 
In-Reply-To: <200311080335.hA83Zh15003385@a-night-at-the-opera.cu.groogr
	oo.com>
References: <3.0.5.32.20031107194446.00fb8b38@pop.ncsa.uiuc.edu>
Message-ID: <3.0.5.32.20031107222430.00fb8b38@pop.ncsa.uiuc.edu>

At 09:35 PM 11/7/2003 -0600, Arun Bhalla wrote:
> Well, I think now it's time to look at the code.  I've never seen Net::RTP
> before.  It looks like rtp_recv is either inefficient or is being
> used incorrectly.  Perhap you can look through the Net::RTP code and
> figure out what's going on.

Hi, Arun....

The Net::RTP code is the RAT (Robust Audio Tool) rtp.c common library
(v1.2.14) run through SWIG to make it available to Perl.  It should be
pretty solid, since it's recent, and it's been pounded on a bunch for a
long time by lots of folks.

However, I'm wondering about my call to it.  I had trouble with the socket
stuff earlier on, where I wasn't getting all the data from the connection
-- This seemed to clear it up, but now I wonder if it's not hanging up
somewhere.  Does this look clean to you?



# Continue to loop until rtp_recv returns FALSE -- Read 
# *all* the data, not just one packet's worth.
# 	Args: What are we listening to, how long can we block, and when did we
get it?
while (rtp_recv($session, $timeout, $rtp_ts)) { 
	rtp_update($session);
}


Thanks again!

Mitch
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From bhalla at uiuc.edu  Fri Nov  7 23:15:57 2003
From: bhalla at uiuc.edu (Arun Bhalla)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Re: Perl Debugger and finding CPU Usage 
In-Reply-To: <3.0.5.32.20031107222430.00fb8b38@pop.ncsa.uiuc.edu> 
Message-ID: <200311080515.hA85Fv15004197@a-night-at-the-opera.cu.groogroo.com>


Mitch Kutzko writes:
> At 09:35 PM 11/7/2003 -0600, Arun Bhalla wrote:
> > Well, I think now it's time to look at the code.  I've never seen Net::RTP
> > before.  It looks like rtp_recv is either inefficient or is being
> > used incorrectly.  Perhap you can look through the Net::RTP code and
> > figure out what's going on.
> 
> Hi, Arun....
> 
> The Net::RTP code is the RAT (Robust Audio Tool) rtp.c common library
> (v1.2.14) run through SWIG to make it available to Perl.  It should be
> pretty solid, since it's recent, and it's been pounded on a bunch for a
> long time by lots of folks.
> 
> However, I'm wondering about my call to it.  I had trouble with the socket
> stuff earlier on, where I wasn't getting all the data from the connection
> -- This seemed to clear it up, but now I wonder if it's not hanging up
> somewhere.  Does this look clean to you?
> 
> 
> 
> # Continue to loop until rtp_recv returns FALSE -- Read 
> # *all* the data, not just one packet's worth.
> # 	Args: What are we listening to, how long can we block, and when did we
> get it?
> while (rtp_recv($session, $timeout, $rtp_ts)) { 
> 	rtp_update($session);
> }

Given that information ;-), my best guess is that the timeout may be too
short, causing for rtp_recv to be called too frequently.  However, it would
appear that rtp_recv is never timing out -- I would assume that rtp_recv
would return false on timeout, and thus the while loop would end.  Also, I
imagine that your client has certain constraints on proper timeouts.  Still,
I suppose it's worth a shot trying to change the timeout, seeing if
it improves performance and doesn't break your code.

Otherwise maybe you need to check other aspects, such as how your
socket/session are initialized and seeing which parameters are set and
which are not.

Arun

From lewart at uiuc.edu  Mon Nov 10 12:33:59 2003
From: lewart at uiuc.edu (Daniel S. Lewart)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Meeting Tue Nov 11 19:00 @ Manzella's in Champaign
Message-ID: <20031110123358.B20140@staff1.cso.uiuc.edu>

Champaign-Urbana Perl Mongers,

Our next meeting shall be:
  * Tue Nov 11 19:00-21:00
  * Manzella's Italian Patio
  * 115 S First St / Champaign
Hey, that's tomorrow night!

I'm happy to see the mailing list being used for something besides
these announcements! :)

We'll be in the non-smoking (NE) section of the restaurant again.
I suppose we can celebrate the release of Perl 5.8.2:
	http://use.perl.org/articles/03/11/07/117238.shtml

Cheers,
Dan
http://cmi.pm.org/

From sigfried at ojctech.com  Fri Nov  7 15:53:20 2003
From: sigfried at ojctech.com (Sigfried Gold)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage
In-Reply-To: <3.0.5.32.20031107153350.03433788@pop.ncsa.uiuc.edu>
References: <3.0.5.32.20031107153350.03433788@pop.ncsa.uiuc.edu>
Message-ID: <20031107215320.GD1202@che.ojctech.com>

i've done this.  the perl profiler is pretty easy to use.

from perldoc perldebug:

          The Perl Profiler

          If you wish to supply an alternative debugger for Perl to
          run, just invoke your script with a colon and a package
          argument given to the -d flag.  The most popular alternative
          debuggers for Perl is the Perl profiler.  Devel::DProf is
          now included with the standard Perl distribution.  To
          profile your Perl program in the file mycode.pl, just type:

              $ perl -d:DProf mycode.pl

          When the script terminates the profiler will dump the
          profile information to a file called tmon.out.  A tool like
          dprofpp, also supplied with the standard Perl distribution,
          can be used to interpret the information in that profile.

* Mitch Kutzko <mitch@ncsa.uiuc.edu> :
> Hi, folks -- I have a question about the Perl debugger, which I've
> essentially never actually used.
> 
> How do I go about using it to find out where CPU usage is occurring?  I've
> got a very large, very complex script that's soaking up *way* too much CPU
> (but not leaking memory -- already checked for that) for what it's doing,
> but I can't figure out where/how it's happening.
> 
> Any ideas on how to get the debugger to show me where all the CPU is going?
> 
> Mitch
> --
> Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
> http://hobbes.ncsa.uiuc.edu/
> _______________________________________________
> Champaign-Urbana mailing list
> Champaign-Urbana@mail.pm.org
> http://mail.pm.org/mailman/listinfo/champaign-urbana

-- 
Sigfried Gold
Home: 212-568-3549
Cell: 718-781-6991
Work: 212-304-7210
http://www.sigfried.org
mailto:sigfried@sigfried.org

From brandon at ojctech.com  Fri Nov  7 15:54:21 2003
From: brandon at ojctech.com (Brandon Bowersox)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage
In-Reply-To: <3.0.5.32.20031107153350.03433788@pop.ncsa.uiuc.edu>
Message-ID: <5.2.0.9.0.20031107155024.02cd1dd8@localhost>

Mitch,

The Perl Profiler is the tool for you and it is distributed with Perl.  See
http://search.cpan.org/~nwclark/perl-5.8.2/ext/Devel/DProf/DProf.pm
http://search.cpan.org/~nwclark/perl-5.8.2/utils/dprofpp.PL

Basically, to create a profile file, run your code like:
perl -d:DProf script.pl (on the command line)
OR
#!/usr/bin/perl -d:DProf (at the top of your script)

Then the dprofpp command will give you a report.  My favorite reports are:
dprofpp tmon.out (the basic report including CPU time)
dprofpp -S tmon.out (merged subroutine tree)

Brandon

--
Brandon Bowersox
brandon@ojctech.com
OJC Technologies, Inc.
www.ojctech.com, 217-278-3933


At 03:33 PM 11/7/2003 -0600, Mitch Kutzko wrote:
>Hi, folks -- I have a question about the Perl debugger, which I've
>essentially never actually used.
>
>How do I go about using it to find out where CPU usage is occurring?  I've
>got a very large, very complex script that's soaking up *way* too much CPU
>(but not leaking memory -- already checked for that) for what it's doing,
>but I can't figure out where/how it's happening.
>
>Any ideas on how to get the debugger to show me where all the CPU is going?
>
>Mitch
>--
>Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
>http://hobbes.ncsa.uiuc.edu/
>_______________________________________________
>Champaign-Urbana mailing list
>Champaign-Urbana@mail.pm.org
>http://mail.pm.org/mailman/listinfo/champaign-urbana


From ryddler at cu-online.com  Tue Nov 11 10:19:19 2003
From: ryddler at cu-online.com (ryddler)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage
In-Reply-To: <5.2.0.9.0.20031107155024.02cd1dd8@localhost>
References: <5.2.0.9.0.20031107155024.02cd1dd8@localhost>
Message-ID: <1068567559.1252.7.camel@bprice2kl.atgf.com>

With all this talk about the profiler, does anyone know how to invoke it
in a CGI environment, say with mod_perl, and more to the point for me,
from within Apache::ASP?

On Fri, 2003-11-07 at 15:54, Brandon Bowersox wrote:
> Mitch,
> 
> The Perl Profiler is the tool for you and it is distributed with Perl.  See
> http://search.cpan.org/~nwclark/perl-5.8.2/ext/Devel/DProf/DProf.pm
> http://search.cpan.org/~nwclark/perl-5.8.2/utils/dprofpp.PL
> 
> Basically, to create a profile file, run your code like:
> perl -d:DProf script.pl (on the command line)
> OR
> #!/usr/bin/perl -d:DProf (at the top of your script)
> 
> Then the dprofpp command will give you a report.  My favorite reports are:
> dprofpp tmon.out (the basic report including CPU time)
> dprofpp -S tmon.out (merged subroutine tree)
> 
> Brandon
> 
> --
> Brandon Bowersox
> brandon@ojctech.com
> OJC Technologies, Inc.
> www.ojctech.com, 217-278-3933
> 
> 
> At 03:33 PM 11/7/2003 -0600, Mitch Kutzko wrote:
> >Hi, folks -- I have a question about the Perl debugger, which I've
> >essentially never actually used.
> >
> >How do I go about using it to find out where CPU usage is occurring?  I've
> >got a very large, very complex script that's soaking up *way* too much CPU
> >(but not leaking memory -- already checked for that) for what it's doing,
> >but I can't figure out where/how it's happening.
> >
> >Any ideas on how to get the debugger to show me where all the CPU is going?
> >
> >Mitch
> >--
> >Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
> >http://hobbes.ncsa.uiuc.edu/
> >_______________________________________________
> >Champaign-Urbana mailing list
> >Champaign-Urbana@mail.pm.org
> >http://mail.pm.org/mailman/listinfo/champaign-urbana
> 
> _______________________________________________
> Champaign-Urbana mailing list
> Champaign-Urbana@mail.pm.org
> http://mail.pm.org/mailman/listinfo/champaign-urbana
-- 
Keep the turbo spooled!
ryddler
http://www.starquestclub.com


From mitch at ncsa.uiuc.edu  Tue Nov 11 11:00:44 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl installable for Win2K/XP?
Message-ID: <3.0.5.32.20031111110044.037723e8@pop.ncsa.uiuc.edu>

Hi, folks -- Another question for you....

I'm looking for a free version of Perl for Win2K/WinXP systems.  Something
recent, stable, and compatable with CPAN, preferably.

What recommendations does anyone have?  The last time I had to solve this
problem, I got lucky and someone had a licensed copy of the MKS Utilities
version of Perl, which is now *way* too old to be useful in this context.

Thanks in advance!

Mitch

PS - Still working on that CPU usage problem -- I'll keep you posted when I
figure it out. ;-)
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From bhalla at uiuc.edu  Tue Nov 11 11:13:58 2003
From: bhalla at uiuc.edu (Arun Bhalla)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl Debugger and finding CPU Usage 
In-Reply-To: <1068567559.1252.7.camel@bprice2kl.atgf.com> 
Message-ID: <200311111713.hABHDw15030427@a-night-at-the-opera.cu.groogroo.com>


I saw Apache::DProf and Apache::SmallProf listed on CPAN, and I can only guess
that they are the mod_perl versions of the same.  Worth checking out.

Arun

ryddler writes:
> With all this talk about the profiler, does anyone know how to invoke it
> in a CGI environment, say with mod_perl, and more to the point for me,
> from within Apache::ASP?
> 
> On Fri, 2003-11-07 at 15:54, Brandon Bowersox wrote:
> > Mitch,
> > 
> > The Perl Profiler is the tool for you and it is distributed with Perl.  See
> > http://search.cpan.org/~nwclark/perl-5.8.2/ext/Devel/DProf/DProf.pm
> > http://search.cpan.org/~nwclark/perl-5.8.2/utils/dprofpp.PL
> > 
> > Basically, to create a profile file, run your code like:
> > perl -d:DProf script.pl (on the command line)
> > OR
> > #!/usr/bin/perl -d:DProf (at the top of your script)
> > 
> > Then the dprofpp command will give you a report.  My favorite reports are:
> > dprofpp tmon.out (the basic report including CPU time)
> > dprofpp -S tmon.out (merged subroutine tree)
> > 
> > Brandon
> > 
> > --
> > Brandon Bowersox
> > brandon@ojctech.com
> > OJC Technologies, Inc.
> > www.ojctech.com, 217-278-3933
> > 
> > 
> > At 03:33 PM 11/7/2003 -0600, Mitch Kutzko wrote:
> > >Hi, folks -- I have a question about the Perl debugger, which I've
> > >essentially never actually used.
> > >
> > >How do I go about using it to find out where CPU usage is occurring?  I've
> > >got a very large, very complex script that's soaking up *way* too much CPU
> > >(but not leaking memory -- already checked for that) for what it's doing,
> > >but I can't figure out where/how it's happening.
> > >
> > >Any ideas on how to get the debugger to show me where all the CPU is going
> ?
> > >
> > >Mitch
> > >--
> > >Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
> > >http://hobbes.ncsa.uiuc.edu/
> > >_______________________________________________
> > >Champaign-Urbana mailing list
> > >Champaign-Urbana@mail.pm.org
> > >http://mail.pm.org/mailman/listinfo/champaign-urbana
> > 
> > _______________________________________________
> > Champaign-Urbana mailing list
> > Champaign-Urbana@mail.pm.org
> > http://mail.pm.org/mailman/listinfo/champaign-urbana
> -- 
> Keep the turbo spooled!
> ryddler
> http://www.starquestclub.com
> 
> _______________________________________________
> Champaign-Urbana mailing list
> Champaign-Urbana@mail.pm.org
> http://mail.pm.org/mailman/listinfo/champaign-urbana
> 

--
Arun Bhalla

From mitch at ncsa.uiuc.edu  Tue Nov 11 11:16:21 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl installable for Win2K/XP?
In-Reply-To: <006301c3a876$e311aea0$0e07a8c0@researchpress.com>
References: <3.0.5.32.20031111110044.037723e8@pop.ncsa.uiuc.edu>
Message-ID: <3.0.5.32.20031111111621.037723e8@pop.ncsa.uiuc.edu>

Not yet, but this:

	http://www.activestate.com/Products/Download/Download.plex?id=ActivePerl

Looks like exactly what I need.  

Thanks!

Mitch

At 11:11 AM 11/11/2003 -0600, you wrote:
> Tried ActiveState's Perl?
> 
> - jeff
> 
> 
> --
> Jeff Helgesen
> http://www.shout.net/~jmh/
> 
> 
> ----- Original Message -----
> From: "Mitch Kutzko" <mitch@ncsa.uiuc.edu>
> To: "CU Perlmongers" <champaign-urbana@mail.pm.org>
> Sent: Tuesday, November 11, 2003 11:00 AM
> Subject: [CMI.PM] Perl installable for Win2K/XP?
> 
> 
> > Hi, folks -- Another question for you....
> >
> > I'm looking for a free version of Perl for Win2K/WinXP systems.  Something
> > recent, stable, and compatable with CPAN, preferably.
> >
> > What recommendations does anyone have?  The last time I had to solve this
> > problem, I got lucky and someone had a licensed copy of the MKS Utilities
> > version of Perl, which is now *way* too old to be useful in this context.
> >
> > Thanks in advance!
> >
> > Mitch
> >
> > PS - Still working on that CPU usage problem -- I'll keep you posted when
> I
> > figure it out. ;-)
> > --
> > Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
> > http://hobbes.ncsa.uiuc.edu/
> > _______________________________________________
> > Champaign-Urbana mailing list
> > Champaign-Urbana@mail.pm.org
> > http://mail.pm.org/mailman/listinfo/champaign-urbana
> >
> 
> 
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From ethomson at ravecomm.com  Tue Nov 11 11:16:30 2003
From: ethomson at ravecomm.com (Ed Thomson)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl installable for Win2K/XP?
In-Reply-To: <3.0.5.32.20031111110044.037723e8@pop.ncsa.uiuc.edu>
References: <3.0.5.32.20031111110044.037723e8@pop.ncsa.uiuc.edu>
Message-ID: <20031111171630.GA23949@mail.ravecomm.com>

Mitch Kutzko wrote:
> Hi, folks -- Another question for you....
> 
> I'm looking for a free version of Perl for Win2K/WinXP systems.  Something
> recent, stable, and compatable with CPAN, preferably.
> 
> What recommendations does anyone have?  The last time I had to solve this
> problem, I got lucky and someone had a licensed copy of the MKS Utilities
> version of Perl, which is now *way* too old to be useful in this context.

I always install cygwin <http://www.cygwin.org/> on windows boxes.
It helps ease the pain of... well, windows.  It includes a recent
version of Perl.  I don't know how well it deals with CPAN, but I've
never had any problems with it.

-Ed

From mcovingt at ux1.cso.uiuc.edu  Tue Nov 11 11:16:42 2003
From: mcovingt at ux1.cso.uiuc.edu (Maiko Covington)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl installable for Win2K/XP?
In-Reply-To: <3.0.5.32.20031111110044.037723e8@pop.ncsa.uiuc.edu>
Message-ID: <Pine.GSO.4.31.0311111113520.28685-100000@staff2.cso.uiuc.edu>

Hiya.

On Tue, 11 Nov 2003, Mitch Kutzko wrote:

> Hi, folks -- Another question for you....
>
> I'm looking for a free version of Perl for Win2K/WinXP systems.  Something
> recent, stable, and compatable with CPAN, preferably.
>
> What recommendations does anyone have?  The last time I had to solve this
> problem, I got lucky and someone had a licensed copy of the MKS Utilities
> version of Perl, which is now *way* too old to be useful in this context.

	I've had good luck with ActiveState perl, at www.activestate.com.
It just installs in the usual windows fashion, and comes with a perl
package manager to get modules.  The version I have is perl 5.8.

	I've found the modules I need there from active state and in the
perl package manager myself, all recent, but YMMV.

	Maiko Covington



From ryddler at cu-online.com  Tue Nov 11 15:04:48 2003
From: ryddler at cu-online.com (ryddler)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Mongering tonight?
Message-ID: <1068584688.1252.41.camel@bprice2kl.atgf.com>

Hey Dan,

Does all this traffic on the list mean we're going to have more people
at the meeting tonight? There is that usage fee, right? ;)

-- 
Keep the turbo spooled!
ryddler
http://www.starquestclub.com


From cclausen at acm.org  Tue Nov 11 15:21:33 2003
From: cclausen at acm.org (Christopher D. Clausen)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Perl installable for Win2K/XP?
References: <3.0.5.32.20031111110044.037723e8@pop.ncsa.uiuc.edu>
	<20031111171630.GA23949@mail.ravecomm.com>
Message-ID: <025301c3a89a$244d9500$8d0f15ac@ad.uiuc.edu>

Ed Thomson <ethomson@ravecomm.com> wrote:
> Mitch Kutzko wrote:
>> Hi, folks -- Another question for you....
>>
>> I'm looking for a free version of Perl for Win2K/WinXP systems.
>> Something recent, stable, and compatable with CPAN, preferably.
>>
>> What recommendations does anyone have?  The last time I had to solve
>> this problem, I got lucky and someone had a licensed copy of the MKS
>> Utilities version of Perl, which is now *way* too old to be useful
>> in this context.
>
> I always install cygwin <http://www.cygwin.org/> on windows boxes.
> It helps ease the pain of... well, windows.  It includes a recent
> version of Perl.  I don't know how well it deals with CPAN, but I've
> never had any problems with it.

I strongly recommend against using cygwin on anything but personal,
single user systems.  The readme states that cygwin uses shared memory
spaces for data storage and there is the possibility that processes can
access memory of another application.

<<CDC
Christopher D. Clausen



From mitch at ncsa.uiuc.edu  Wed Nov 12 14:45:21 2003
From: mitch at ncsa.uiuc.edu (Mitch Kutzko)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Authoritative hostname?
Message-ID: <3.0.5.32.20031112144521.0106cde8@pop.ncsa.uiuc.edu>

Hi again!

Different question today:  What's the best way you recommend to reliably
get an authoritative Fully Qualified Domain Name for a given machine?

I've been using gethostbyaddr() and doing a reverse lookup on the IP
address, but sometimes that comes back with "woof" instead of
"woof.uiuc.edu" if the /etc/hosts file isn't set up correctly.

On the machine itself, I can use the Net::Domain::getfqdn() call, but that
sometimes gives spurious results depending on the contents of
/etc/resolv.conf.  And in any event, if other folks are doing a
gethostbyaddr(), the name they each come back with for the same machine may
not agree with each other.

So.  Any thoughts?

Mitch
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/

From lewart at uiuc.edu  Wed Nov 12 14:49:24 2003
From: lewart at uiuc.edu (Daniel S. Lewart)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] Authoritative hostname?
In-Reply-To: <3.0.5.32.20031112144521.0106cde8@pop.ncsa.uiuc.edu>;
	from mitch@ncsa.uiuc.edu on Wed, Nov 12, 2003 at 02:45:21PM -0600
References: <3.0.5.32.20031112144521.0106cde8@pop.ncsa.uiuc.edu>
Message-ID: <20031112144923.A14063@staff1.cso.uiuc.edu>

Mitch,

> Different question today:  What's the best way you recommend to reliably
> get an authoritative Fully Qualified Domain Name for a given machine?

perldoc -q hostname

Cheers,
Dan
http://cmi.pm.org/

From brandon at ojctech.com  Sat Nov 15 13:48:34 2003
From: brandon at ojctech.com (Brandon Bowersox)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] block STDERR from a CPAN module implemented in C?
Message-ID: <20031115194834.GA21042@che.ojctech.com>

Since we've had such lively support lately...

How can I trap or block STDERR output from the XML::GDOME module 
which is an XS module that uses a C implementation?

Pretend Solution #1------------------------------------------
close STDERR;
XML::GDOME->foo; #call that generates STDERR

Pretend Solution #2------------------------------------------
local *STDERR;
open (STDERR, ">".File::Spec->devnull());
XML::GDOME->foo; #call that generates STDERR

#1 works but stops me from getting any of the helpful output from any 
part of my program below.  #2 does not work because the XML::GDOME 
calls are in a different scope where "local *STDERR" no longer 
applies.

Why do I want to do this?  XML::GDOME gives "critical" error messages 
on STDERR even when it works correctly and produces the right output.
The implementation seems to have trouble with reference counting and 
it hates when, for example, a document object is garbage collected 
when there are still node objects of that doc that have not been 
garbage collected.  XML::GDOME is no longer being actively improved, 
but I want to use it because it works and supports the DOM methods 
I need, including importNode().  And I want to capture the STDERR 
so all these messages do not fill up or overwhelm the Apache mod_perl 
CGI environment from which I'm calling XML::GDOME.

Brandon Bowersox

From mepstein at uiuc.edu  Sat Nov 15 18:16:44 2003
From: mepstein at uiuc.edu (Milt Epstein)
Date: Mon Aug  2 21:27:36 2004
Subject: [CMI.PM] block STDERR from a CPAN module implemented in C?
In-Reply-To: <20031115194834.GA21042@che.ojctech.com>
Message-ID: <Pine.GSO.4.31.0311151813350.20891-100000@staff3.cso.uiuc.edu>

On Sat, 15 Nov 2003, Brandon Bowersox wrote:

> Since we've had such lively support lately...
>
> How can I trap or block STDERR output from the XML::GDOME module
> which is an XS module that uses a C implementation?
>
> Pretend Solution #1------------------------------------------
> close STDERR;
> XML::GDOME->foo; #call that generates STDERR
>
> Pretend Solution #2------------------------------------------
> local *STDERR;
> open (STDERR, ">".File::Spec->devnull());
> XML::GDOME->foo; #call that generates STDERR
>
> #1 works but stops me from getting any of the helpful output from any
> part of my program below.  #2 does not work because the XML::GDOME
> calls are in a different scope where "local *STDERR" no longer
> applies.
[ ... ]

Did you try it without the "local *STDERR"?  (You can save and restore
it before and after as needed -- there's an example on the perlfunc
man page.)

Milt Epstein
Research Programmer
Integration and Software Engineering (ISE)
Campus Information Technologies and Educational Services (CITES)
University of Illinois at Urbana-Champaign (UIUC)
mepstein@uiuc.edu