Monitor total run time of processes

[mbianchy]

Dear fellow monitoring people,

I wonder if (to be precise: how) I could monitor the total run time of a process per time frame.

Let's say I have a process which I can monitor by proc_info[] and collect data like memory consumption etc.
The process in question is running at different times of day and can run not at all for a couple of days as well as several time per day with interruptions. Given the resolution of collecting samples (once per minute) it is not 100% sure to exactly tell if the process is running like just 10 seconds and is always "off" if the agent is "looking for it". Usually the process runs for a minimum of 20..30 minutes per run

My question is: Can I tell the total run time of the process per day? i.e. the process was running in total 2h35m yesterday, 1h2m today etc?

I tried a trigger with an expression min(/hostname/proc_info[my_shiny_process.exe,wkset,sum],1h)>0 and did experiments with different time frames (like 24h) but the trigger clears as soon the process is finished and doesn't like to tell what I'd like to know.
Creating some scripts "around" the process in question is no option - and the platform of the agent is Windoze while the server is running on Linux - but the brain grease is the trigger independent from where the values are coming from?

Somebody points me in the proper direction, please?

[mbianchy]

Stupid me - after thinking a while about totally different things I noticed my question is already containing the answer: I just use a cheap count() function for the number of samples where the value is non zero - that's it. Cheap as in "no fancy calculation necessary" and not in "no credit card needed".

I virtually haven't seen the wood for the trees.

The item will be a calculated item with a formula like this:
count(//my_process.exe,wkset,sum],24h)>0