>>Does anyone know how you go about determining what causes regular, long
>>pauses like this?
>
> I'd start by using a profiler.

>Could a P/Invoke call
>create a lot of temporary objects behind the scenes?

>Does anyone know how you go about determining what causes regular, long
>pauses like this?

"Charles Jenkins" <not.my.email.address@not.here.com> wrote in message
news:Xns999A618254E6Fcharlesedwardjenkins@207.46.248.16...
>I have a C# application that calls a lower-level C++ dll to do serial
> communication. (Using C++ for serial I/O is not necessary, but I was asked
> to do this.)
>
> Calls from C# to C++ functions generally look like this:
>
> [DllImport( DLL_NAME, EntryPoint = "SetMultiplier" )]
> private static extern WFGP_MSG DllSetMultiplier(
> Int16 channel, UInt16 mult
> );
>
> All the calls made during a communication session use types that require
> no
> translation in order to cross the boundary between managed and unmanaged
> code. The P/Invoke calls should be just as fast as P/Invoke can be.
>
> Everything works just fine. However, the application tends to "freeze" for
> a few seconds after every few minutes of use. No data gets lost, so the
> freeze is more of an annoyance than anything. However, I feel sure that it
> is caused by something I am not doing correctly in the way I use unmanaged
> code. My first instinct would be do assume that regular, long pauses are
> caused by creating many, many objects that almost immediately go out of
> scope, leaving a huge amount of work for the garbage collector. However, I
> do not believe my code is creating tons of objects. Could a P/Invoke call
> create a lot of temporary objects behind the scenes?
>
> Does anyone know how you go about determining what causes regular, long
> pauses like this?