opensta-devel — OpenSTA developer discussions

[OpenSTA-devel] [ opensta-Bugs-1406872 ] SEED for REPEATABLE RANDOM variables is ignored

[SourceForge.net <no...@so...>]

Bugs item #1406872, was opened at 2006-01-15 17:05
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110857&aid=1406872&group_id=10857

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: Script Language
Group: Behavioral
Status: Open
Resolution: None
Priority: 5
Submitted By: Daniel Sutcliffe (dansut)
Assigned to: Daniel Sutcliffe (dansut)
Summary: SEED for REPEATABLE RANDOM variables is ignored

Initial Comment:
The ability to specifiy a SEED value to variables
specified as REPEATABLE RANDOM is intended to give the
possibility of different repeatable psuedo random
sequences being attainable from the same set of
possible results.  In the OpenSTA 1.4.3 release, and
probably all other 1.x releases up to this one, the
specified SEED is ignored completely and the same
repeatable sequence is created whatever value is specified.

There is no real workaround for this issue.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110857&aid=1406872&group_id=10857

RE: [OpenSTA-devel] Re: Platform SDK

[Wickersham, P. <Pet...@di...>]

Ok, I'm in the midst of building CVS HEAD with no STLPort, VC 6 SP6,
Platform SDK Feb 2003 edition, everything else standard.

I'm running into some compilation errors with some of the STL classes.
Do you have these resolved already? If so, I'll wait for any of your
checkins.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Thursday, January 12, 2006 5:49 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: Platform SDK

Peter Wickersham wrote:
> Ok, I have finally built the entire source, created an installation
> kit, installed and run a test.

Congratulations!  I have a suspicion that not many people make it
this far ... it isn't an easy road :-(  The goal has to be to make
it easier.
=20
> I ended up ditching the most recent Platform SDK and going with the
> February 2003 Platform SDK edition since that is what MSDN said was
> supported for VC 6.0.  I did that before your last email on working
> around the linking issue, so I didn't give that a try.=20

This could actually be the right decision to make, and the
recommended build environment for the 1.4.4 release.  I think that is
the Platform SDK that Thierry was using to some success,  I chose to
go with the "latest" platform SDK simply due to the fact I knew it
was going to be easily attainable.  If the Feb 2003 Platform SDK is
going to be easily available in the long term and this is the last
PSDK that is recommended for VC6 then there is a strong argument that
it is the one we should be recommending for the 1.4 series - given
goood long term testing results.

> I did use STLPort 4.6.2 with the _STLP_NEW_PLATFORM_SDK defined in
> the stl_user_config.h header file.=20

When you previously said "the latest" STLPort I thought you were
going for STLPort 5.  The 5 release seems to be completely
unpublished on the stlport.org Web site - if you go to the STLPort
SourceForge project site you'll find that 5.0.1 was recently released
and is actually supposed to be the latest ... I haven't tried this
release though and all my recent testing suggests that OpenSTA really
has no need to use STLPort, at least for the windows only 1.4 series,
in fact using STLPort actually makes some parts of OpenSTA much
slower.

I'd really like to see you using the CVS HEAD and no STLPort ... ;-)

> Other than the Platform SDK and STLPort, I used all the other
> recommended versions.  This is with VC 6.0 SP6.  The only issue
> with STLPort that I saw was in OmniOrb in 3 files WRT iostreams.
> In the end, I just directly modified the OmniOrb source to do the
> standard includes and namespace declaration. The 3 files were:
>   appl/omniNames/log.h
>   appl/omniNames/log.cc
>   appl/omniMapper/omniMapper.cc

Now you mention it, that is just what I vaguely remember having to do
when I was still experimenting with the STLPort.

> The only other change that I recall had to do with the installation
> project because I had to add the new version of the STL dll.

Yes, this DLL changed name as I remember.  Getting rid of it is the
way forward :-)

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

Re: [OpenSTA-devel] Re: When is Random not random?

[Mark E. <ma...@bo...>]

Wickersham, Peter wrote:
> Ok, so now that I can build, I have added a random number generator
> class based on the Mersenne Twister PRNG. It is fast, uniform, and has a
> long period (iterations before repeating).
> 
> However, I have run into issues in actually getting the seed from the
> script set in the code. I reviewed the current code in CECRangeVar and
> CECListVar and it looked like those classes were always ignoring the
> seed and just using a random one to start. So, I decided to test this,
> but I can't seem to get any SEED I set in the script to actually get
> passed into the constuctors for these variable classes. They always get
> 0, which is the default higher in the stack if no seed is specified. 
> 
> Does anyone know the SCL parsing/compilation pieces enough to help me
> track down where the SEED seems to get lost.
> 
try the file varlist.cxx

CECVar * Cvarlist::define_variable(const string& scriptname, const
TRecord * rec, const string& dir, bool use_decl)
...
has the lines
     // Extract seed for random values
     long seed;
     par = rec->param("seed");
     if(par == NULL)
         seed = 0;
     else
         seed = par->intVal();
to extract the seed from the TOF

however SCL seem to be generating an ivalue object not "seed"
GenericObjects.odl seem to be expecting an ivalue so the above code
should probably change the literal "seed" to the constant
TParam::stIvalue.

If that still doesn't work I can look a bit further.

Mark

[OpenSTA-devel] [RE] Update Notification

[Tami M. <ba...@ci...>]

Further Increase In Tourist Numbers Expected with Completion of the Western 3rd Round 
City Highway, Connecting to Qin E-Pang Palace


Company: Dark Dynamite Inc.
Symbol: DKDY
Thursday Close: $1.86
Short Term: $2.25 - $2.50
Note: New Press Release
Indi cator: S T R O N G   B U Y 


DKDY seems to be catching all the breaks. With the announcement of their new nightly 
Banquets, Dance and Cultural show begining in April 2006, and now with the annoucement 
a new High Way will be completed reducing drive time and ease for tourists to reach the 
park.

This is expected to increase the number of visitors due to convenience of access to the 
park, and thus increase annual revenues even further.

The last few days for Dark Dynamite INc. (DKDY) have been exciting. With large increasees 
in volume and price jumps to $2.10 from the last news release, we expect we have not seen 
the last of it with this new info coming to the table.

Members we are puting DKDY on the watch list again for tomorrow. This company has really 
sparked our interest as not only an exciting day trade but an excelent long term 
investment as well.

For those of you who got in at the end of the day when the price dipped, congrats, 
for those of you who didn't, get on it first thing before the price jumps again over $2.00.

Just read the news and see for yourself the promising future we see for DKDY

The News:

XI'AN, China, Jan. 4 /Xinhua-PRNewswire/ -- Dark Dynamite Inc. (OTC Bulletin Board: DKDY) 
(''DDI'') announces that the construction of the Western 3rd Round City Highway (the ''3rd 
Highway'') in Xi'an is expected to be completed in early 2006. The completed 3rd Highway is 
expected to bring in more tourists to the Theme Park of the Qin E-Pang Palace 
(''E-Pang Palace'').

The 3rd Highway is invested in by the Xi'an Government, with a total investment of RMB500 
million (equal to US$61.7 million). The total length of the 3rd Highway is 18.42 km, which 
includes 7 large crossroads and 2 bridges. The 3rd Highway is expected to be completed early 
2006, and to be opened free of charge. After the completion of the 3rd Round City Highway, 
traveling time between the city center and the Theme Park of E-pang Palace will be reduced 
from 30 minutes to around 20 minutes, thus greatly benefiting tourists for traveling to the 
E-Pang Palace, as well as promoting the attractiveness of the E-Pang Palace for tourists.

Since the is Theme Park located at the centre of many historic sites of Xi'an city, such as 
the famous Emperor Qin's Terra-cotta Warriors and Horses Museum, the Mausoleum of Emperor 
Qin, Huaqing Hot-springs, and Fa'men Temple. After the completion of the 3rd Highway, 
tourists will have easy access to the Theme Park while visiting other historical sites. 
The Theme Park is estimated to become one of the main tourism destinations among travel 
agencies due to its distinctive Qin cultural palaces and convenient traffic conditions, 
which is estimated to bring an extra 10,000 tourists, as well as an increase in income.

The Xi'an tourism market saw 650,000 overseas tourists, and 20.84 million domestic tourists 
traveling to Xi'an in 2004, with increase rates of 93% and 29.1% respectively comparing with 
those of the previous year. The total revenue of the Xi'an tourism industry was 
RMB15.44 billion (equal to US$1.9 billion) in 2004, which includes foreign exchange of 
US$330 million, with respective increase rates of 45.1% and 130%.


DKDY, thats our pick.

Good Trading Members

[OpenSTA-devel] Re: Codemax

[Federico T. <ft...@fi...>]

> That is the normal/good way to validate a Web transaction...
>
> However, often doing this causes problems because specific Web
> applications may be very dynamic in the HTML they create and trying
> to match to a DOM identifier will fail even though the transaction
> was successful.
>
> The alternative is to not use a DOM identifier, to dump the whole
> body, and then to use ~LOCATE, ~EXTRACT, etc. to pull out your text.
> The problem with this can be that you are limited to the first 64K of
> content...
>

Perfect, an other good issue, but it is not what I want.

Let me say that the system that I tested was made with autogenerate code.
It has a lot of HTTP trafic and parameters. Some times, I made chages
to the original script to parametrize, but this didn't work. Or
perhaps it works fine to one VU but not with 2 or more VU.
Using BView we can debug a lot of the script running alone. I want to
a facilitie to debug a script that has problems to run with two or
more VU.
I want to see the pages, complete html pages, that the server return
with error (aplication errors), that dont pass the test case.


> > > The Modeler can already use an ODBC data source to populate an
> > > FVR file - is this the area you would like to extend?  Or would
> > > you want to integrate into the Commander, or just be another
> > > seperate tool?
> >=3D20
> > ok, but the functionality that you say in the Modeler populate an
> > FVR at the moment that we create the variable.  After an execution
> > we can not update the contenent of a FVR quickly and easily.
>
> OK, I think the "right" initial way forward is to expand the
> Modeler's existing facilities to make it possible to update the FVR
> contents at any time in the most flexible manner.
>

I'd like to expand the Commander with this facilitie, because I want
to update all the variables existing in all the scripts contained in a
Test. Previusly we could set a couple of parameters, and then just
with a click update all the FVRs correspondig.


> Warning: there are bugs associated with the whole FVR updating and
> usage area.  Check the bug database.

I know the bugs, and I had to use the workarrounds

>
> > Now I use my tool as a separate tool, but could be good that all be
> > together in the same toolset, perhaps as a part of the Commander.
>
> An important part of any features you provide in this area should be
> a command line interface.  Many people run their tests in "batch
> mode" where they are not interested in the GUI at all - this type
> of usage is a key element of OpenSTAs future goals.
>

Ok, thanks

saludos!
fedeFede

[OpenSTA-devel] Re: Codemax

[Daniel S. <da...@Op...>]

Daniel Sutcliffe wrote:
> > OK, so this is the key question: these "application errors", what
> > signifies them?  Because, as far as I am concerned there is no
> > standard way of reporting Web application errors - HTTP error
> > codes certainly normally have little relation to Web application
> > errors.

Federico Toledo wrote:
> In my case, for example, I want to receive a page with the message
> "Facturaci=F3n completa" so I use a Load response info ... and compare
> the string.

That is the normal/good way to validate a Web transaction...

However, often doing this causes problems because specific Web
applications may be very dynamic in the HTML they create and trying
to match to a DOM identifier will fail even though the transaction
was successful.

The alternative is to not use a DOM identifier, to dump the whole
body, and then to use ~LOCATE, ~EXTRACT, etc. to pull out your text.
The problem with this can be that you are limited to the first 64K of
content...

> But if the load response info fails because not match with the page
> I can't know what happened.  I could save a log...

This is the code fragment I provided earlier.

> but I want to save the HTML page in the case that fail the load
> response info ... or fail the test case, and then, after of a
> execution, I could see the pages returned in each error.

The real problem you face with writing out a complete file after a=20
script action is that of doing file IO from SCL.  The problems that
exist in this area are mostly related to the potential distributed
nature of Script execution.

> This facilitates the concurrence test more than the stress test.
> Ok?

I think it is acknowledged that even in stress/load testing you need
to know at what point things start to fail ...

> Understand my intention?

I totally understand your intention, and think I always have.  I'm
still not sure of what you would like to implement though.

> > The Modeler can already use an ODBC data source to populate an
> > FVR file - is this the area you would like to extend?  Or would
> > you want to integrate into the Commander, or just be another
> > seperate tool?
>=20
> ok, but the functionality that you say in the Modeler populate an
> FVR at the moment that we create the variable.  After an execution
> we can not update the contenent of a FVR quickly and easily.

OK, I think the "right" initial way forward is to expand the
Modeler's existing facilities to make it possible to update the FVR
contents at any time in the most flexible manner.

Warning: there are bugs associated with the whole FVR updating and
usage area.  Check the bug database.

> Now I use my tool as a separate tool, but could be good that all be
> together in the same toolset, perhaps as a part of the Commander.

An important part of any features you provide in this area should be
a command line interface.  Many people run their tests in "batch
mode" where they are not interested in the GUI at all - this type
of usage is a key element of OpenSTAs future goals.

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

[OpenSTA-devel] Re: Platform SDK

[Daniel S. <da...@tc...>]

Peter Wickersham wrote:
> Ok, I have finally built the entire source, created an installation
> kit, installed and run a test.

Congratulations!  I have a suspicion that not many people make it
this far ... it isn't an easy road :-(  The goal has to be to make
it easier.
 
> I ended up ditching the most recent Platform SDK and going with the
> February 2003 Platform SDK edition since that is what MSDN said was
> supported for VC 6.0.  I did that before your last email on working
> around the linking issue, so I didn't give that a try. 

This could actually be the right decision to make, and the
recommended build environment for the 1.4.4 release.  I think that is
the Platform SDK that Thierry was using to some success,  I chose to
go with the "latest" platform SDK simply due to the fact I knew it
was going to be easily attainable.  If the Feb 2003 Platform SDK is
going to be easily available in the long term and this is the last
PSDK that is recommended for VC6 then there is a strong argument that
it is the one we should be recommending for the 1.4 series - given
goood long term testing results.

> I did use STLPort 4.6.2 with the _STLP_NEW_PLATFORM_SDK defined in
> the stl_user_config.h header file. 

When you previously said "the latest" STLPort I thought you were
going for STLPort 5.  The 5 release seems to be completely
unpublished on the stlport.org Web site - if you go to the STLPort
SourceForge project site you'll find that 5.0.1 was recently released
and is actually supposed to be the latest ... I haven't tried this
release though and all my recent testing suggests that OpenSTA really
has no need to use STLPort, at least for the windows only 1.4 series,
in fact using STLPort actually makes some parts of OpenSTA much
slower.

I'd really like to see you using the CVS HEAD and no STLPort ... ;-)

> Other than the Platform SDK and STLPort, I used all the other
> recommended versions.  This is with VC 6.0 SP6.  The only issue
> with STLPort that I saw was in OmniOrb in 3 files WRT iostreams.
> In the end, I just directly modified the OmniOrb source to do the
> standard includes and namespace declaration. The 3 files were:
>   appl/omniNames/log.h
>   appl/omniNames/log.cc
>   appl/omniMapper/omniMapper.cc

Now you mention it, that is just what I vaguely remember having to do
when I was still experimenting with the STLPort.

> The only other change that I recall had to do with the installation
> project because I had to add the new version of the STL dll.

Yes, this DLL changed name as I remember.  Getting rid of it is the
way forward :-)

Cheers
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

RE: [OpenSTA-devel] Re: When is Random not random?

[Wickersham, P. <Pet...@di...>]

Ok, so now that I can build, I have added a random number generator
class based on the Mersenne Twister PRNG. It is fast, uniform, and has a
long period (iterations before repeating).

However, I have run into issues in actually getting the seed from the
script set in the code. I reviewed the current code in CECRangeVar and
CECListVar and it looked like those classes were always ignoring the
seed and just using a random one to start. So, I decided to test this,
but I can't seem to get any SEED I set in the script to actually get
passed into the constuctors for these variable classes. They always get
0, which is the default higher in the stack if no seed is specified.=20

Does anyone know the SCL parsing/compilation pieces enough to help me
track down where the SEED seems to get lost.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of
Wickersham, Peter
Sent: Tuesday, January 10, 2006 2:10 PM
To: ope...@li...
Subject: RE: [OpenSTA-devel] Re: When is Random not random?


I agree with Dan, the issue is not the window between srand() and rand()
calls, it is that fact the Windows rand() doesn't actually use its' last
result as the new seed. So, the call to srand() is messing up the
uniformity of the number distribution. Not to mention that rand() is not
32-bit, which is another problem.

So, I have been trying to get things built to take a look at a new
generator class that is 32-bit, allows for multiple generators and is
highly uniform.=20

My issue is that I only have Visual Studio 2003 at the moment. I decided
to give this a try, but have found that omniOrb 3.0.4 doesn't compile
with VS 2003, so I decided to try omniOrb 4.0.6, which does. Since I am
using VS 2003, I don't need the Platform SDK. However, I keep running
into little issues along the way. The most recent being a difference
between omniOrb 3 and 4 WRT to connections and thread pools. I guess
thread pools were a new feature in omniOrb 4.

At this point, I'm looking for VS 6 and plan on going back to the basic
components for the 1.4.3 source.=20

I'll keep you all posted.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Sunday, January 08, 2006 6:45 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: When is Random not random?

Mark Elam wrote:
> Yes, but we're calling srand(A1),rand(),srand(a2),rand(),srand(a3)
> ...  So the 'initialize with seed' is occuring before each call to
> rand, which should reset the seed before each call.  However if
> another thread called srand() before the first thread rand() had
> occured I can see the problem you are describing which is why I
> wondered whether just making the srand(),rand() pair atomic (with a
> mutex) would solve the issue.  As rand() would always be called
> after a reset with srand() using the same seed this should give the
> same sequence.

I think you are correct Mark.  I've tried to stimulate this potential
concurrency and timing issue to occur using some contrived scripts
and I can't seem to - that's not to say it won't happen, just that it
seems it is unlikely to happen.  Maybe it's something to do with how
the work is assigned to the threads in a Test Executer, or the
potential points of context switch on my single CPU test host.

The above doesn't matter though, as the whole method of always
reseeding is obviously a really bad way of generating "random"
numbers...  I created a little script like this:
  Definitions
    CHARACTER*512 USER_AGENT
    INTEGER USE_PAGE_TIMERS
    INTEGER RandInt(1-10), RANDOM
    INTEGER Occurrence[10]
    INTEGER Idx
  Code
    Entry[USER_AGENT,USE_PAGE_TIMERS]
    DO Idx =3D 1, 1000
        GENERATE RandInt
        SET Occurrence[RandInt] =3D Occurrence[RandInt] + 1
    ENDDO
    DO Idx =3D 1, 10
       LOG Idx, " occurred ", Occurrence[Idx], " times"
    ENDDO
Running this, very simple, test several times was enough to tell me
that there is definitely not an equal probabiliy of getting each of
the possible returns of RandInt.  I found by, unscientifically,
eyeing the results that there was always much more chance of getting
a 1, 3, or 8; and much less of a chance of getting 2, 5, or 7... not
good!

Obviously the behavior of RANDOM variables needs fixing - it looks
like there's 3 seperate bugs that can be addressed in one update:
 - limited range of returns (bug#415144)
 - less than random distribution of returns (not logged)
 - small chance of REPEATABLE failing due to timing (not logged)
I'm still concerned about keeping the old (bad) behavior of
REPEATABLE as default, with a logged warning, that can then have the
good behavior, or warning, defeated using the .ini file.  My feeling
is that it is not worthwhile keeping the behavior of regular RANDOM
but other opinions are welcome.

I'm starting to put a shortlist together for a 1.4.4 release and it
would be great if Peter (or someone) could donate some code to get
these problems, at least partially, fixed for this release.

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_idv37&alloc_id=16865&op=3Dick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

RE: [OpenSTA-devel] Re: When is Random not random?

[Wickersham, P. <Pet...@di...>]

Ok, so now that I can build, I have added a random number generator
class based on the Mersenne Twister PRNG. It is fast, uniform, and has a
long period (iterations before repeating).

However, I have run into issues in actually getting the seed from the
script set in the code. I reviewed the current code in CECRangeVar and
CECListVar and it looked like those classes were always ignoring the
seed and just using a random one to start. So, I decided to test this,
but I can't seem to get any SEED I set in the script to actually get
passed into the constuctors for these variable classes. They always get
0, which is the default higher in the stack if no seed is specified.=20

Does anyone know the SCL parsing/compilation pieces enough to help me
track down where the SEED seems to get lost.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of
Wickersham, Peter
Sent: Tuesday, January 10, 2006 2:10 PM
To: ope...@li...
Subject: RE: [OpenSTA-devel] Re: When is Random not random?


I agree with Dan, the issue is not the window between srand() and rand()
calls, it is that fact the Windows rand() doesn't actually use its' last
result as the new seed. So, the call to srand() is messing up the
uniformity of the number distribution. Not to mention that rand() is not
32-bit, which is another problem.

So, I have been trying to get things built to take a look at a new
generator class that is 32-bit, allows for multiple generators and is
highly uniform.=20

My issue is that I only have Visual Studio 2003 at the moment. I decided
to give this a try, but have found that omniOrb 3.0.4 doesn't compile
with VS 2003, so I decided to try omniOrb 4.0.6, which does. Since I am
using VS 2003, I don't need the Platform SDK. However, I keep running
into little issues along the way. The most recent being a difference
between omniOrb 3 and 4 WRT to connections and thread pools. I guess
thread pools were a new feature in omniOrb 4.

At this point, I'm looking for VS 6 and plan on going back to the basic
components for the 1.4.3 source.=20

I'll keep you all posted.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Sunday, January 08, 2006 6:45 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: When is Random not random?

Mark Elam wrote:
> Yes, but we're calling srand(A1),rand(),srand(a2),rand(),srand(a3)
> ...  So the 'initialize with seed' is occuring before each call to
> rand, which should reset the seed before each call.  However if
> another thread called srand() before the first thread rand() had
> occured I can see the problem you are describing which is why I
> wondered whether just making the srand(),rand() pair atomic (with a
> mutex) would solve the issue.  As rand() would always be called
> after a reset with srand() using the same seed this should give the
> same sequence.

I think you are correct Mark.  I've tried to stimulate this potential
concurrency and timing issue to occur using some contrived scripts
and I can't seem to - that's not to say it won't happen, just that it
seems it is unlikely to happen.  Maybe it's something to do with how
the work is assigned to the threads in a Test Executer, or the
potential points of context switch on my single CPU test host.

The above doesn't matter though, as the whole method of always
reseeding is obviously a really bad way of generating "random"
numbers...  I created a little script like this:
  Definitions
    CHARACTER*512 USER_AGENT
    INTEGER USE_PAGE_TIMERS
    INTEGER RandInt(1-10), RANDOM
    INTEGER Occurrence[10]
    INTEGER Idx
  Code
    Entry[USER_AGENT,USE_PAGE_TIMERS]
    DO Idx =3D 1, 1000
        GENERATE RandInt
        SET Occurrence[RandInt] =3D Occurrence[RandInt] + 1
    ENDDO
    DO Idx =3D 1, 10
       LOG Idx, " occurred ", Occurrence[Idx], " times"
    ENDDO
Running this, very simple, test several times was enough to tell me
that there is definitely not an equal probabiliy of getting each of
the possible returns of RandInt.  I found by, unscientifically,
eyeing the results that there was always much more chance of getting
a 1, 3, or 8; and much less of a chance of getting 2, 5, or 7... not
good!

Obviously the behavior of RANDOM variables needs fixing - it looks
like there's 3 seperate bugs that can be addressed in one update:
 - limited range of returns (bug#415144)
 - less than random distribution of returns (not logged)
 - small chance of REPEATABLE failing due to timing (not logged)
I'm still concerned about keeping the old (bad) behavior of
REPEATABLE as default, with a logged warning, that can then have the
good behavior, or warning, defeated using the .ini file.  My feeling
is that it is not worthwhile keeping the behavior of regular RANDOM
but other opinions are welcome.

I'm starting to put a shortlist together for a 1.4.4 release and it
would be great if Peter (or someone) could donate some code to get
these problems, at least partially, fixed for this release.

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_idv37&alloc_id=16865&op=3Dick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

[OpenSTA-devel] Re: Codemax

[Federico T. <ft...@fi...>]

>
> OK, so this is the key question: these "application errors", what
> signifies them?  Because, as far as I am concerned there is no
> standard way of reporting Web application errors - HTTP error codes
> certainly normally have little relation to Web application errors.
>

In my case, for example, i want to receive a page with the message
"Facturaci=F3n completa" so I use a Load response info ... and compare
the string.
But if the load response info fails because not match with the page i
can't know what happened. I could save a log...but i want to save the
HTML page in the case that fail the load response info ... or fail the
test case, and then, after of a execution, i could see the pages
returned in each error.
This facilitates the concurrence test more than the stress test. Ok?
Understand my intention?


> > I supouse that you like me parameterize your scripts, and get your
> > values from file.
> > This files you have to generate with some data from DataBase with
> > the corresponding SQL.
> > I have 10 .fvr with parameters.  After each execution I have to
> > refresh the contenent of this 10 archives.  My tool gets the SQL
> > from a file (acording how you configure a XML), and execute this
> > and save the result in the corresponding .fvr.
> > Before I made the tool, I take 10 minutes to do this job, and now I
> > take 1 second to press the buton.
> > I made this and anothers tools in java, but could be better that we
> > can integrate all of this to OpenSTA.
>
> The Modeler can already use an ODBC data source to populate an FVR
> file - is this the area you would like to extend?  Or would you want
> to integrate into the Commander, or just be another seperate tool?

ok, but the functionality that you say in the Modeler populate an FVR
at the moment that we create the variable. After an execution we can
not update the contenent of a FVR quickly and easily.
Now i use my tool as a separate tool, but could be good that all be
together in the same toolset, perhaps as a part of the Commander.

saludos!
FedeFede

RE: [OpenSTA-devel] Re: Platform SDK

[Wickersham, P. <Pet...@di...>]

Ok, I have finally built the entire source, created an installation kit,
installed and run a test.

I ended up ditching the most recent Platform SDK and going with the
February 2003 Platform SDK edition since that is what MSDN said was
supported for VC 6.0. I did that before your last email on working
around the linking issue, so I didn't give that a try.=20

I did use STLPort 4.6.2 with the _STLP_NEW_PLATFORM_SDK defined in the
stl_user_config.h header file.=20

Other than the Platform SDK and STLPort, I used all the other
recommended versions. This is with VC 6.0 SP6. The only issue with
STLPort that I saw was in OmniOrb in 3 files WRT iostreams. In the end,
I just directly modified the OmniOrb source to do the standard includes
and namespace declaration. The 3 files were:

appl/omniNames/log.h
appl/omniNames/log.cc
appl/omniMapper/omniMapper.cc

The only other change that I recall had to do with the installation
project because I had to add the new version of the STL dll.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Wednesday, January 11, 2006 2:31 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: Platform SDK

Peter Wickersham wrote:
> I went ahead and grabbed the latest Platform SDK (you provided a
> link) and the latest STLPort.

Using the latest STLport could be trouble - although I haven't tried
doing this in a while now and maybe they've fixed some things.
Please report back on this.

There's a mail from me in the archives from back last June 19th that
discusses a few of the issues I was having.  I'd love to post a link
but the SF archives seem to be "sleeping" at the moment.  The same
mail mentions the below issue ...

> However, in reading the SDK notes, it specifically says that VC 6.0
> is not compatible with the SDK libraries.  If you try and link with
> them you get errors.  And, in fact, that is exactly what happened
> with me on the Commander project.
> Has someone here built with this combination?
>=20
> --------------------Configuration: Commander - Win32
Release--------------------
> Linking...
> htmlhelp.lib(init.obj) : error LNK2001: unresolved external symbol
___security_cookie
> htmlhelp.lib(init.obj) : error LNK2001: unresolved external symbol
@__security_check_cookie@4
> ..\bin\Release\OSCommander.exe : fatal error LNK1120: 2 unresolved
externals
> Error executing link.exe.
> OSCommander.exe - 3 error(s), 0 warning(s)

It just took me an age to work out what I did to fix this ...
What you need to do is make the earlier HtmlHelp library and include
dirs come before the Platform SDK dirs in your VS6 "Options >
Directories" dialog.  The earlier HtmlHelp stuff should come from
your "HTML Help Workshop" installation.  At least that is what I did.

This link has another idea which I haven't tried:
  http://blogs.msdn.com/nikolad/archive/2005/01/27/362214.aspx

Keep us informed with how you get on,
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

[OpenSTA-devel] Re: Codemax

[Daniel S. <da...@Op...>]

Federico Toledo wrote:
> I will explain a little more.
> In some cases, in a execution I see that the server return
> aplication's errors and so the test case fail.

OK, so this is the key question: these "application errors", what
signifies them?  Because, as far as I am concerned there is no
standard way of reporting Web application errors - HTTP error codes
certainly normally have little relation to Web application errors.

> If I want to know why it fails, what try to say me the aplication,
> I can't.  For example, to do this to one VU I use BView, because I
> can view the pages returned by the aplication.  It's alright.  But
> if I run the script with two VU and there are problems, because the
> aplication fails with concurrents users (for example too), I'd like
> to view the pages with error to detect the error.  The easiest form
> to do this is that OpenSTA save this pages automatically.  I hope
> that you understand me.

I think I understand what your intentions are, I just can't
understand how you intend to spot the error automatically.

> > Of Course you can do something like this now by saying:
> >   LOAD RESPONSE_INFO BODY ON ConId INTO VarName, &
> >     WITH DomIdentifier, RETURNING STATUS RetStat
> >   IF (RetStat < 0) THEN
> >     LOAD RESPONSE_INFO BODY ON ConId INTO VarName
> >     LOG VarName
> >   ENDIF
> > but perhaps something more elegant and automatic would be much
> > more useful.
> 
> Ok, that's a good idea, that could help me with other problems, but
> it isn't what I want.

If you want to do something that it is not caused by contents of the
returned HTTP Reponse body; say you want to make your condition on
the HTTP return code then you just need a ",RETURNING CODE HttpCode"
on your SCL HTTP request command, then change the conditional based
on this return.  If you did base it on HTTP error code then for many
codes the returned BODY would not be at all helpful.  Does that help
you more?

If not, give us more of a clue of what the condition you would base
your "failure log" on.

> I supouse that you like me parameterize your scripts, and get your
> values from file.
> This files you have to generate with some data from DataBase with
> the corresponding SQL.
> I have 10 .fvr with parameters.  After each execution I have to
> refresh the contenent of this 10 archives.  My tool gets the SQL
> from a file (acording how you configure a XML), and execute this
> and save the result in the corresponding .fvr.
> Before I made the tool, I take 10 minutes to do this job, and now I
> take 1 second to press the buton.
> I made this and anothers tools in java, but could be better that we
> can integrate all of this to OpenSTA.

The Modeler can already use an ODBC data source to populate an FVR
file - is this the area you would like to extend?  Or would you want
to integrate into the Commander, or just be another seperate tool?

Cheers
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

[OpenSTA-devel] Re: Platform SDK

[Daniel S. <da...@op...>]

Daniel Sutcliffe wrote:
> There's a mail from me in the archives from back last June 19th
> that discusses a few of the issues I was having.  I'd love to post
> a link but the SF archives seem to be "sleeping" at the moment.

Here we go:
  http://sourceforge.net/mailarchive/message.php?msg_id=12110026

/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org

[OpenSTA-devel] Re: Platform SDK

[Daniel S. <da...@Op...>]

Peter Wickersham wrote:
> I went ahead and grabbed the latest Platform SDK (you provided a
> link) and the latest STLPort.

Using the latest STLport could be trouble - although I haven't tried
doing this in a while now and maybe they've fixed some things.
Please report back on this.

There's a mail from me in the archives from back last June 19th that
discusses a few of the issues I was having.  I'd love to post a link
but the SF archives seem to be "sleeping" at the moment.  The same
mail mentions the below issue ...

> However, in reading the SDK notes, it specifically says that VC 6.0
> is not compatible with the SDK libraries.  If you try and link with
> them you get errors.  And, in fact, that is exactly what happened
> with me on the Commander project.
> Has someone here built with this combination?
> 
> --------------------Configuration: Commander - Win32 Release--------------------
> Linking...
> htmlhelp.lib(init.obj) : error LNK2001: unresolved external symbol ___security_cookie
> htmlhelp.lib(init.obj) : error LNK2001: unresolved external symbol @__security_check_cookie@4
> ..\bin\Release\OSCommander.exe : fatal error LNK1120: 2 unresolved externals
> Error executing link.exe.
> OSCommander.exe - 3 error(s), 0 warning(s)

It just took me an age to work out what I did to fix this ...
What you need to do is make the earlier HtmlHelp library and include
dirs come before the Platform SDK dirs in your VS6 "Options >
Directories" dialog.  The earlier HtmlHelp stuff should come from
your "HTML Help Workshop" installation.  At least that is what I did.

This link has another idea which I haven't tried:
  http://blogs.msdn.com/nikolad/archive/2005/01/27/362214.aspx

Keep us informed with how you get on,
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org

Re:[OpenSTA-devel] Re: Re: Codemax

[Federico T. <ft...@fi...>]

Ok, thanks for your comments (all)

> I'd like to request a slightly different feature (if I understand you
> correctly): Saving a page should be explicitly triggered by the script
> whenever the script would like to save a response - not on HTTP errors
> in general: I believe that load tests can easily and correctly test that
> state 404 is returned by the application. On the other hand if the page
> is returned with http-state 200 but contains unexpected content (e.g. an
> application error message) the script should be able to save this
> unexpected content.

> Presumably you have done some sort of conditional before your:
>   FAIL TEST-CASE
> Why not just LOG the error (or BODY contents) before doing the fail.

> How are the "test cases" defined?
> Do the pages get written out each to their own file, or in a log?

> If you just want to do it by HTTP Code then this would be relatively
> easily done in SCL - except if you want them writing out to seperate
> files, or if they are longer than 64K...

I will explain a little more.
In some cases, in a execution I see that the server return
aplication's errors and so the test case fail. If I want to know why
it fails, what try to say me the aplication, i can't.
For example, to do this to one VU I use BView, because I can view the
pages returned by the aplication. It's allright. But if I run the
script with two VU and there are problems, because the aplication
fails with concurrents users (for example too), I'd like to view the
pages with error to detect the error.
The easiest form to do this is that OpenSTA save this pages automatically.
I hope that you understand me.


> Your description makes me think of one worthwhile change that could
> probably be made fairly easily and is unlikely to upset any existing
> users:
>
> If after doing a GET/POST you try to pull something from the BODY
> content returned using:
>   LOAD RESPONSE_INFO BODY ON ConId INTO VarName, WITH DomIdentifier
>
> And the DomIdentifier does not match the structure of the content
> returned, then perhaps it would be nicer (rather than falling on its
> face with "unresolved variable" as it does now) if the whole BODY
> could be logged somewhere or placed in a variable.
>
> Of Course you can do something like this now by saying:
>   LOAD RESPONSE_INFO BODY ON ConId INTO VarName, &
>     WITH DomIdentifier, RETURNING STATUS RetStat
>   IF (RetStat < 0) THEN
>     LOAD RESPONSE_INFO BODY ON ConId INTO VarName
>     LOG VarName
>   ENDIF
> but perhaps something more elegant and automatic would be much more
> useful.

Ok, thats a good idea, that could helpme with other problems, but it
isn't what I want.



> > Then I want to investigate how to work with SOA and WebService, but
> > this will be in the future.
>
> I think you'll find lots of people interested in this type of work
> and ideas that make testing this type of use of HTTP easier or more
> complete will be most welcomed.

Ok, thats true, but for now I dont know a lot about so I'll try on
this in the feature.



> > I have implement a couple of tools in Java to do some activities
> > that can be done automaticaly.
> > For example, a tool to refresh the script's parameters files
> > (because after runing a test there are data that is no more valid,
> > so we have to repeat the SQL and export the result to the script's
> > parameters files to can run again)
> > I would like to integrate this funcionality to OpenSTA.
>
> The ability to run "tools" that can return your test system to its
> baseline standard either at the end or prior to starting a test, is a
> problem that many people have solved in many different ways.  Ideas
> of ways to make these tasks generically easier to automate within
> OpenSTAs toolset would be welcomed.

I supouse that you like me parametrice your scripts, and get your
values from file.
This files you have to generate with some data from DataBase with the
corresponding SQL.
I have 10 .fvr with parameters. After each execution I have to refresh
the contenent of this 10 archives. My tool gets the sql from a file
(acording how you configure a XML), and execute this and save the
result in the corresponding .fvr.
Before i made the tool, I take 10 minutes to do this job, and now I
take 1 second to press the buton.
I made this and anothers tools in java, but could be better that we
can integrate all of this to openSTA.


thks to all
greetings
FedeFede

RE: [OpenSTA-devel] Re: Platform SDK

[Wickersham, P. <Pet...@di...>]

Ah, I must have missed the bit about what changes were in the HEAD
earlier. I'll probably move back to it once I get the hang of the build
with the 1.4.3 source since I'm already along the way there.

I went ahead and grabbed the latest Platform SDK (you provided a link)
and the latest STLPort. However, in reading the SDK notes, it
specifically says that VC 6.0 is not compatible with the SDK libraries.
If you try and link with them you get errors. And, in fact, that is
exactly what happened with me on the Commander project. Has someone here
built with this combination?

--------------------Configuration: Commander - Win32
Release--------------------
Linking...
htmlhelp.lib(init.obj) : error LNK2001: unresolved external symbol
___security_cookie
htmlhelp.lib(init.obj) : error LNK2001: unresolved external symbol
@__security_check_cookie@4
..\bin\Release\OSCommander.exe : fatal error LNK1120: 2 unresolved
externals
Error executing link.exe.

OSCommander.exe - 3 error(s), 0 warning(s)

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Tuesday, January 10, 2006 5:15 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: Platform SDK

Wickersham, Peter wrote:
> So, I have the 1.4.3 source, omniOrb 3.0.4, ucd-snmp 4.26, CodeMax,
> SizeCBar 2.44, and I dug up VS 6.0 with SP6.
>=20
> The only real missing piece for me was Platform SDK. I'll give the
> newest a try and see what I run into.

There shouldn't be anything to worry about building 1.4.3 source
package - you don't mention STLport here though and if you are going
to build with this you really need to make sure this is set up first
so that everything gets built using it also.

> Don't worry about fast tracking a commit to the HEAD since I am now
> focused on the 1.4.3 release.

Whether you decide to build with or without the STLport you should
still use the CVS HEAD over the 1.4.3 source package - if you
already have CVS installed then this is a no-brainer.  As I said in
another reply: the only differences between the CVS HEAD and the
1.4.3 source package are currently some useful bug fixes - no changes
in build environment yet.

When I do check my changes in I will only check the updates that will
work with or without the STLport into the HEAD.  The changes that
force you not to use the STLport will go in a branch to be merged
back into the trunk once a few other poeple have verified them.

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

[OpenSTA-devel] Re: Platform SDK

[Daniel S. <da...@Op...>]

Karl Hoffmann wrote:
> I think you can choose an 2000 enviroment in the 2003 SP 1 SDK

You certainly can, but that's not what we're talking about when we
say the 2000 Platform SDK.  What is actually referred to is the
Platform SDK the Microsoft released in July 2000, with all its bugs
and peculiarities ;-)

Selecting the build environment from the 2003 Platform SDK menus
simply sets up environment variables for library/include PATHs, etc.
And you shouldn't notice any difference (problems) with OpenSTA if
you select any of the 32 bit retail environments.  Don't ask me what
happens if you select any of the 64 bit environments - I don't know,
but I bet it isn't pretty :-)

Cheers
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

Re: [OpenSTA-devel] Re: Platform SDK

[karl h. <k.h...@og...>]

Hello *,

I think you can choose an 2000 enviroment in the 2003 SP 1 SDK

start menu, chose Microsoft Platform SDK February 2003->Open Build 
Environment Window->Windows 2000 Build Environment->Windows 2000 Build 
Environment (Retail),

karl

Wickersham, Peter schrieb:
> Ok.. Well, I was fighting CVS HEAD with just Visual Studio 2003, no
> Platform SDK, no STLPort, omniOrb 4.0.6 and a newer SNMP. Besides all
> the library references to OmniOrb, the biggest compilation issues seem
> to come from VS 2003 differences.
> 
> At this point, I am now in the middle of setting up a more default
> environment against 1.4.3.
> 
> So, I have the 1.4.3 source, omniOrb 3.0.4, ucd-snmp 4.26, CodeMax,
> SizeCBar 2.44, and I dug up VS 6.0 with SP6.
> 
> The only real missing piece for me was Platform SDK. I'll give the
> newest a try and see what I run into. Don't worry about fast tracking a
> commit to the HEAD since I am now focused on the 1.4.3 release.
> 
> thanks!
> peter
> 
> -----Original Message-----
> From: ope...@li...
> [mailto:ope...@li...] On Behalf Of Daniel
> Sutcliffe
> Sent: Tuesday, January 10, 2006 4:44 PM
> To: ope...@li...
> Subject: [OpenSTA-devel] Re: Platform SDK
> 
> Wickersham, Peter wrote:
>> I am having quite a time finding a 2000 version of the SDK.  MSDN
>> online only offers the 2001 versions and up at this point.  Anyone
>> have any luck with those, or should I continue my search for a 2000
>> MSDN CD?
> 
> It's been a long time since I've any luck finding it online. :-(
> 
> I think we're at the point where we can safely say that using the
> Windows 2003 Server SP1 Platform SDK is OK:
>  
> http://www.microsoft.com/downloads/details.aspx?FamilyId=A55B6B43-E24F-4
> EA3-A93E-40C0EC4F68E5&displaylang=en
> 
> I'm not sure about building OmniORB 3.2.5 with this though - I seem
> to remember having to make some small fix ... If I find it in my
> notes I'll post it here.
> 
> You probably shouldn't try OmniORB 4 though - Mark Elam reported
> that although it can be made to build, it doesn't work for some
> reason he hasn't sussed yet... unless he has an update?
> 
> Are you building from the CVS HEAD?  If so I'll put some effort
> tomorrow into getting the stuff commited so that you don't need to
> fight with STLport ...
> 
> Cheers
> /dan

-- 
*************************************************************
ogvit gmbh & co kg     * wap.ogvit.de      * adresse
internet-technologien  * fon 05251.687060  * jesuitenmauer 24
www.ogvit.de           * fax 05251.687069  * 33098 paderborn
*************************************************************
c/o Postalo gmbh
kattrepel 2 - 20095 hamburg
tel. 040/51318373  - fax 040/51318374

ogvit ist ein IBM business Partner
*************************************************************

[OpenSTA-devel] Re: Platform SDK

[Daniel S. <da...@Op...>]

Wickersham, Peter wrote:
> So, I have the 1.4.3 source, omniOrb 3.0.4, ucd-snmp 4.26, CodeMax,
> SizeCBar 2.44, and I dug up VS 6.0 with SP6.
> 
> The only real missing piece for me was Platform SDK. I'll give the
> newest a try and see what I run into.

There shouldn't be anything to worry about building 1.4.3 source
package - you don't mention STLport here though and if you are going
to build with this you really need to make sure this is set up first
so that everything gets built using it also.

> Don't worry about fast tracking a commit to the HEAD since I am now
> focused on the 1.4.3 release.

Whether you decide to build with or without the STLport you should
still use the CVS HEAD over the 1.4.3 source package - if you
already have CVS installed then this is a no-brainer.  As I said in
another reply: the only differences between the CVS HEAD and the
1.4.3 source package are currently some useful bug fixes - no changes
in build environment yet.

When I do check my changes in I will only check the updates that will
work with or without the STLport into the HEAD.  The changes that
force you not to use the STLport will go in a branch to be merged
back into the trunk once a few other poeple have verified them.

Cheers
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

RE: [OpenSTA-devel] Re: Platform SDK

[Wickersham, P. <Pet...@di...>]

Ok.. Well, I was fighting CVS HEAD with just Visual Studio 2003, no
Platform SDK, no STLPort, omniOrb 4.0.6 and a newer SNMP. Besides all
the library references to OmniOrb, the biggest compilation issues seem
to come from VS 2003 differences.

At this point, I am now in the middle of setting up a more default
environment against 1.4.3.

So, I have the 1.4.3 source, omniOrb 3.0.4, ucd-snmp 4.26, CodeMax,
SizeCBar 2.44, and I dug up VS 6.0 with SP6.

The only real missing piece for me was Platform SDK. I'll give the
newest a try and see what I run into. Don't worry about fast tracking a
commit to the HEAD since I am now focused on the 1.4.3 release.

thanks!
peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Tuesday, January 10, 2006 4:44 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: Platform SDK

Wickersham, Peter wrote:
> I am having quite a time finding a 2000 version of the SDK.  MSDN
> online only offers the 2001 versions and up at this point.  Anyone
> have any luck with those, or should I continue my search for a 2000
> MSDN CD?

It's been a long time since I've any luck finding it online. :-(

I think we're at the point where we can safely say that using the
Windows 2003 Server SP1 Platform SDK is OK:
=20
http://www.microsoft.com/downloads/details.aspx?FamilyId=3DA55B6B43-E24F-=
4
EA3-A93E-40C0EC4F68E5&displaylang=3Den

I'm not sure about building OmniORB 3.2.5 with this though - I seem
to remember having to make some small fix ... If I find it in my
notes I'll post it here.

You probably shouldn't try OmniORB 4 though - Mark Elam reported
that although it can be made to build, it doesn't work for some
reason he hasn't sussed yet... unless he has an update?

Are you building from the CVS HEAD?  If so I'll put some effort
tomorrow into getting the stuff commited so that you don't need to
fight with STLport ...

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

[OpenSTA-devel] Re: Platform SDK

[Daniel S. <da...@Op...>]

Wickersham, Peter wrote:
> I am having quite a time finding a 2000 version of the SDK.  MSDN
> online only offers the 2001 versions and up at this point.  Anyone
> have any luck with those, or should I continue my search for a 2000
> MSDN CD?

It's been a long time since I've any luck finding it online. :-(

I think we're at the point where we can safely say that using the
Windows 2003 Server SP1 Platform SDK is OK:
  http://www.microsoft.com/downloads/details.aspx?FamilyId=A55B6B43-E24F-4EA3-A93E-40C0EC4F68E5&displaylang=en

I'm not sure about building OmniORB 3.2.5 with this though - I seem
to remember having to make some small fix ... If I find it in my
notes I'll post it here.

You probably shouldn't try OmniORB 4 though - Mark Elam reported
that although it can be made to build, it doesn't work for some
reason he hasn't sussed yet... unless he has an update?

Are you building from the CVS HEAD?  If so I'll put some effort
tomorrow into getting the stuff commited so that you don't need to
fight with STLport ...

Cheers
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

[OpenSTA-devel] Platform SDK

[Wickersham, P. <Pet...@di...>]

I am having quite a time finding a 2000 version of the SDK. MSDN online
only offers the 2001 versions and up at this point. Anyone have any luck
with those, or should I continue my search for a 2000 MSDN CD?

-peter=20

RE: [OpenSTA-devel] Re: When is Random not random?

[Wickersham, P. <Pet...@di...>]

I agree with Dan, the issue is not the window between srand() and rand()
calls, it is that fact the Windows rand() doesn't actually use its' last
result as the new seed. So, the call to srand() is messing up the
uniformity of the number distribution. Not to mention that rand() is not
32-bit, which is another problem.

So, I have been trying to get things built to take a look at a new
generator class that is 32-bit, allows for multiple generators and is
highly uniform.=20

My issue is that I only have Visual Studio 2003 at the moment. I decided
to give this a try, but have found that omniOrb 3.0.4 doesn't compile
with VS 2003, so I decided to try omniOrb 4.0.6, which does. Since I am
using VS 2003, I don't need the Platform SDK. However, I keep running
into little issues along the way. The most recent being a difference
between omniOrb 3 and 4 WRT to connections and thread pools. I guess
thread pools were a new feature in omniOrb 4.

At this point, I'm looking for VS 6 and plan on going back to the basic
components for the 1.4.3 source.=20

I'll keep you all posted.

-peter

-----Original Message-----
From: ope...@li...
[mailto:ope...@li...] On Behalf Of Daniel
Sutcliffe
Sent: Sunday, January 08, 2006 6:45 PM
To: ope...@li...
Subject: [OpenSTA-devel] Re: When is Random not random?

Mark Elam wrote:
> Yes, but we're calling srand(A1),rand(),srand(a2),rand(),srand(a3)
> ...  So the 'initialize with seed' is occuring before each call to
> rand, which should reset the seed before each call.  However if
> another thread called srand() before the first thread rand() had
> occured I can see the problem you are describing which is why I
> wondered whether just making the srand(),rand() pair atomic (with a
> mutex) would solve the issue.  As rand() would always be called
> after a reset with srand() using the same seed this should give the
> same sequence.

I think you are correct Mark.  I've tried to stimulate this potential
concurrency and timing issue to occur using some contrived scripts
and I can't seem to - that's not to say it won't happen, just that it
seems it is unlikely to happen.  Maybe it's something to do with how
the work is assigned to the threads in a Test Executer, or the
potential points of context switch on my single CPU test host.

The above doesn't matter though, as the whole method of always
reseeding is obviously a really bad way of generating "random"
numbers...  I created a little script like this:
  Definitions
    CHARACTER*512 USER_AGENT
    INTEGER USE_PAGE_TIMERS
    INTEGER RandInt(1-10), RANDOM
    INTEGER Occurrence[10]
    INTEGER Idx
  Code
    Entry[USER_AGENT,USE_PAGE_TIMERS]
    DO Idx =3D 1, 1000
        GENERATE RandInt
        SET Occurrence[RandInt] =3D Occurrence[RandInt] + 1
    ENDDO
    DO Idx =3D 1, 10
       LOG Idx, " occurred ", Occurrence[Idx], " times"
    ENDDO
Running this, very simple, test several times was enough to tell me
that there is definitely not an equal probabiliy of getting each of
the possible returns of RandInt.  I found by, unscientifically,
eyeing the results that there was always much more chance of getting
a 1, 3, or 8; and much less of a chance of getting 2, 5, or 7... not
good!

Obviously the behavior of RANDOM variables needs fixing - it looks
like there's 3 seperate bugs that can be addressed in one update:
 - limited range of returns (bug#415144)
 - less than random distribution of returns (not logged)
 - small chance of REPEATABLE failing due to timing (not logged)
I'm still concerned about keeping the old (bad) behavior of
REPEATABLE as default, with a logged warning, that can then have the
good behavior, or warning, defeated using the .ini file.  My feeling
is that it is not worthwhile keeping the behavior of regular RANDOM
but other opinions are welcome.

I'm starting to put a shortlist together for a 1.4.4 release and it
would be great if Peter (or someone) could donate some code to get
these problems, at least partially, fixed for this release.

Cheers
/dan
--=20
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log
files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
_______________________________________________
OpenSTA-devel mailing list
Ope...@li...
https://lists.sourceforge.net/lists/listinfo/opensta-devel

[OpenSTA-devel] Re: Codemax

[Daniel S. <da...@Op...>]

Olaf Kock wrote:
> I'd like to request a slightly different feature (if I understand
> you correctly): Saving a page should be explicitly triggered by
> the script whenever the script would like to save a response - not
> on HTTP errors in general: I believe that load tests can easily
> and correctly test that state 404 is returned by the application.
> On the other hand if the page is returned with http-state 200 but
> contains unexpected content (e.g. an application error message) the
> script should be able to save this unexpected content.

Your description makes me think of one worthwhile change that could
probably be made fairly easily and is unlikely to upset any existing
users:

If after doing a GET/POST you try to pull something from the BODY
content returned using:
  LOAD RESPONSE_INFO BODY ON ConId INTO VarName, WITH DomIdentifier

And the DomIdentifier does not match the structure of the content
returned, then perhaps it would be nicer (rather than falling on its
face with "unresolved variable" as it does now) if the whole BODY
could be logged somewhere or placed in a variable.

Of Course you can do something like this now by saying:
  LOAD RESPONSE_INFO BODY ON ConId INTO VarName, &
    WITH DomIdentifier, RETURNING STATUS RetStat
  IF (RetStat < 0) THEN
    LOAD RESPONSE_INFO BODY ON ConId INTO VarName
    LOG VarName
  ENDIF
but perhaps something more elegant and automatic would be much more
useful.

Just thinking out loud,
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/

[OpenSTA-devel] Re: Codemax

[Daniel S. <da...@Op...>]

Federico Toledo wrote:
> When I use OpenSTA and run a TaskGroup with validations in the
> scripts (using tests case sentence) if a validation fails I didn't
> know why fails.

Presumably you have done some sort of conditional before your:
  FAIL TEST-CASE
Why not just LOG the error (or BODY contents) before doing the fail.
Maybe I'm missing something but hopefully you can explain further.

> I know that LoadRunner has a facility to know what the server
> answered.  This tool save the pages that didn't pass the test case.
> So we can view the page returned by the server and then we can
> debug the error.

How are the "test cases" defined?
Do the pages get written out each to their own file, or in a log?

> Resuming: I want to save the pages returned with error by the
> server to know why fails.

If you just want to do it by HTTP Code then this would be relatively
easily done in SCL - except if you want them writing out to seperate
files, or if they are longer than 64K...

> Then I want to investigate how to work with SOA and WebService, but
> this will be in the future.

I think you'll find lots of people interested in this type of work
and ideas that make testing this type of use of HTTP easier or more
complete will be most welcomed.

> I have implement a couple of tools in Java to do some activities
> that can be done automaticaly.
> For example, a tool to refresh the script's parameters files
> (because after runing a test there are data that is no more valid,
> so we have to repeat the SQL and export the result to the script's
> parameters files to can run again)
> I would like to integrate this funcionality to OpenSTA.

The ability to run "tools" that can return your test system to its
baseline standard either at the end or prior to starting a test, is a
problem that many people have solved in many different ways.  Ideas
of ways to make these tasks generically easier to automate within
OpenSTAs toolset would be welcomed.

> I have to take a time to do all this 'cause I have to get Visual
> Studio.  I have a question: Could I use dotnet?

To the best of my knowledge, no you can not use .Net or any Visual
Studio version newer than 6 to build OpenSTA.  The projects long term
goal is to move away from requiring ANY development environment that
you have to pay for to build OpenSTAs tools.

Hope this helps,
/dan
-- 
Daniel Sutcliffe <Da...@Op...>
  OpenSTA part-time caretaker - http://OpenSTA.org/