ASP Page Counter

[Chris Hohmann]

Responses inline.

Chris Hohmann wrote on 27 aug 2008 in
microsoft.public.inetserver.asp.general:


One would need to know what they mean by "not need" in this case, Chris.

I assume they meant that no error is thrown if you attempt to modify the
application object without using lock/unlock. There's a contact link at the
bottom of the page if you'd like to request clarification.

How can a state ever be inconsistent?

Anthony already gave an example of inconsistent state.

And I doubt the documentation writers of MS always know what past
implementors did manage to put in the code. Remember the still present
"endif" allowance bug in VBS single line if-else-then?

No way to know for sure unless you ask. I don't remember the still present
"endif" allowance.

The next interesting Q is, how we could ever test this inconsistency,
and more so prove the possible absense of such inconsistency.

Use the hit count example without the lock/unlock and send 100 simultaneous
requests.

 

[Evertjan.]

Chris Hohmann wrote on 27 aug 2008 in
microsoft.public.inetserver.asp.general:

I assume they meant that no error is thrown if you attempt to modify
the application object without using lock/unlock. There's a contact
link at the bottom of the page if you'd like to request clarification.


Anthony already gave an example of inconsistent state.

A state is a state, [because a computer is a state machine].
so it can never be inconsistent.

No way to know for sure unless you ask.

No, in such an organisation, no one will know what happened years before.

I don't remember the still present "endif" allowance.

"We" talked about it a few? years ago on usenet:

if bool then do() end if

should give an error because it should have been just:

if bool then do()

as "end if" is only defined for the [younger] multiline if statement:

if bool then
do()
end if

And is was said by MS that the error was recognized and corrected before,
and then the great corporation websites protested, as their sites had to
much "inconsistent" ;-) pages to change them overnight.

So the error was reintroduced.

Use the hit count example without the lock/unlock and send 100
simultaneous requests.

Simultaneous? How would you do that?

From one machine, they would necessarily not be simultaneous,
and for 100 machines, well, I doubt you could synchronize with some
precision.

And if no strange things happen, as I expect also in the case of absesne
of implied lock, because you cannot do the simultaneous job. would that
prove anything?

 

[Chris Hohmann]

Chris Hohmann wrote on 27 aug 2008 in
microsoft.public.inetserver.asp.general:

I assume they meant that no error is thrown if you attempt to modify
the application object without using lock/unlock. There's a contact
link at the bottom of the page if you'd like to request clarification.


Anthony already gave an example of inconsistent state.

A state is a state, [because a computer is a state machine].
so it can never be inconsistent.

Just because you say it doesn't make it so.

No, in such an organisation, no one will know what happened years before.

You are making a blind assertion here. Have you interviewed every single
person Microsoft?

I don't remember the still present "endif" allowance.

"We" talked about it a few? years ago on usenet:

if bool then do() end if

should give an error because it should have been just:

if bool then do()

as "end if" is only defined for the [younger] multiline if statement:

if bool then
do()
end if

And is was said by MS that the error was recognized and corrected before,
and then the great corporation websites protested, as their sites had to
much "inconsistent" ;-) pages to change them overnight.

So the error was reintroduced.

None of the above changes the fact that I do not recall this discussion. And
more to the point, I do not understand what the argument is. Are you saying
that one error in the ASP documentation/code calls into question ASP in it's
entirety?

Simultaneous? How would you do that?

From one machine, they would necessarily not be simultaneous,
and for 100 machines, well, I doubt you could synchronize with some
precision.

Why don't you try it and let us know how it goes?

 

[Anthony Jones]

Why don't you try it and let us know how it goes?


No.

You could create a repro with just two machines. Have both hammer the
server in large but finite loop. The resulting accumulated count in the
application object should be the sum of loops for both machines. It would
be interesting to see how large the loop would need to be before an
inconsistency would become clear. Or how many additional machines would be
needed.

I grant though in such a simplistic test, conditions may still consipire to
serialise the requests.

(Actually such a test could be produced on a single machine using WinHTTP to
simulate several machines).

I've actually given up trying to convince Evertjan, this sort of thread
hasn't cropped in quite sometime, I'm disappointed that is has now.

 

[Evertjan.]

Anthony Jones wrote on 27 aug 2008 in
microsoft.public.inetserver.asp.general:

I've actually given up trying to convince Evertjan, this sort of thread
hasn't cropped in quite sometime, I'm disappointed that is has now.

Antony,

Why disappointed ?

Are you disappointed not being able to convince someone due to non
convincing arguments?

All the time I stated that you are probably right in your assertion,
but that does not mean it is a fact, as you claimed.

That is why it is theoretically interesting to conceive a test.

In practice it is unimportant, as the exact number of user hits will be
contaminated by the number of bot/crawler hits that no one wants to include
in such count.

Possibly you [general you] can show in such a test, that events are lost,
but the opposite, that no event counts can be lost seems unreachable.

I doubt you [general you] can get enough semiconcurrent page requests where
there is no loss due to a sort of denial of service to test it reliably.

The only way would be reseaching and analyzing the source code. Seems an
hypothetical possiblility.

 

[Chris Hohmann]

Anthony Jones wrote on 27 aug 2008 in
microsoft.public.inetserver.asp.general:


Antony,

Why disappointed ?

Are you disappointed not being able to convince someone due to non
convincing arguments?

All the time I stated that you are probably right in your assertion,
but that does not mean it is a fact, as you claimed.

"All the time" -> hyperbole

That is why it is theoretically interesting to conceive a test.

Unsupported statement of fact.

In practice it is unimportant, as the exact number of user hits will be
contaminated by the number of bot/crawler hits that no one wants to
include
in such count.

Unsupported statement of fact.

Possibly you [general you] can show in such a test, that events are lost,
but the opposite, that no event counts can be lost seems unreachable.

Says you.

I doubt you [general you] can get enough semiconcurrent page requests
where
there is no loss due to a sort of denial of service to test it reliably.

Says you.

The only way would be reseaching and analyzing the source code. Seems an
hypothetical possiblility.

Hyperbole and unsupported statement of fact.

Wouldn't it be easier to admit you were wrong and move on?

 

[Evertjan.]

Chris Hohmann wrote on 28 aug 2008 in
microsoft.public.inetserver.asp.general:

Says you. [..]

Says you. [..]

Hyperbole and unsupported statement of fact.

[please do not quote signatures on usenet]

Wouldn't it be easier to admit you were wrong and move on?

Says you. Sorry I take that back, such childish words.

Wrong in what?

What is wrong with you guys?

I do not state facts, I state lack of facts.

I state a hypothesis, that it is not easy and probably not possible to test
an asp behavour by the suggested method.

But please move on, I am done with you two on this subject.

 

[Chris Hohmann]

But please move on, I am done with you two on this subject.

I win.

 

[Evertjan.]

Chris Hohmann wrote on 28 aug 2008 in
microsoft.public.inetserver.asp.general:

I win.

My childeren loved that too, when they were young.
I am glad to have been of service.

Have a wonderful weekend.

 

[Chris Hohmann]

Chris Hohmann wrote on 28 aug 2008 in
microsoft.public.inetserver.asp.general:


My childeren loved that too, when they were young.
I am glad to have been of service.

Have a wonderful weekend.

Thank you. I had a lovely weekend. I hope you and your family had a pleasant
weekend as well. I'm glad to see you haven't given up on this thread.
Looking forward to your reply.

 

[Evertjan.]

Chris Hohmann wrote on 02 sep 2008 in
microsoft.public.inetserver.asp.general:

Thank you. I had a lovely weekend. I hope you and your family had a
pleasant weekend as well. I'm glad to see you haven't given up on this
thread. Looking forward to your reply.

Well we were quite busy, I am a voluntary guide, and had a large group of
people that as children had been in hiding during the German occupation.
They are elderly now, but have their regular trips together.

And I did some programming and we skyped with our children in two different
continents [from ours].

Time flies like an arrow, and we like this time.

 

[Chris Hohmann]

Chris Hohmann wrote on 02 sep 2008 in
microsoft.public.inetserver.asp.general:

Thank you. I had a lovely weekend. I hope you and your family had a
pleasant weekend as well. I'm glad to see you haven't given up on this
thread. Looking forward to your reply.

Well we were quite busy, I am a voluntary guide, and had a large group of
people that as children had been in hiding during the German occupation.
They are elderly now, but have their regular trips together.

And I did some programming and we skyped with our children in two
different
continents [from ours].

Time flies like an arrow, and we like this time.

Please start a new thread for off-topic posts. Preferably in an appropriate
newsgroup.

 

[Phillip Windell]

Chris Hohmann wrote on 27 aug 2008 in
microsoft.public.inetserver.asp.general:


One would need to know what they mean by "not need" in this case, Chris.
How can a state ever be inconsistent?

Hi guys!
I'm not expert but let me see if I can answer that question by what the
documentation meant. If anything it will help me know if I am following this
correctly.

"not need" = means the value can be altered even if the application is not
locked [implied: even though it is a bad idea]

resulting "inconsistant" = two processes, unaware or each other due to the
absents of a lock, can set the value causing an inconsistancy. For example
each process add "1" to the value, but the final resulting value can end up
incremented by 1 from the original value rather than being incremented by
1,...and then be incremented by 1 again against the resulting value of the
first incrementation. The inconsistancy is that you have a final value of
"1" over the original value instead of "2" over the original value.

Basically the same problem of two users writing to the same field in the
same record of a poorly designed database that results in either incorrect
values in the field, or at worst a corrupt databse?

--
Phillip Windell
www.wandtv.com

The views expressed, are my own and not those of my employer, or Microsoft,
or anyone else associated with me, including my cats.
-----------------------------------------------------