[Mono-list] RT Os and Mono

Lorenzo Viola info at eulogika.net
Fri Feb 15 13:08:16 EST 2008


Martin Olsson ha scritto:
> I have no experience of that stuff but in general; any allocation can 
> trigger full GC and it's fairly hard to predict how long a full GC 
> could take. Allocations is not just caused by "new" etc but also by 
> implicit boxing etc. The latest version of the CLR has a latency mode 
> for the GC but at any rate, writing actual hard limit real-time apps 
> with mono is not a good idea.
>
>
Hello Martin

I admit that I'm quite a newbie about real time, so I've asked this 
question =)

I see that you point out the GC , are you speaking about Garbage 
Collection ?
It was the same first problem I was thinking about =P

I saw that for example on Mono+ASP.NET the GC could be quite intensive,
and not forceable (at last on some month ago version)

btw , what do you mean for implicit boxing ? is it a sort of unwanted 
recursion ?

But IF I work by these terms :

- if the CPU and memory are fast and big enough (like being 10 times the
ususally needed cpu and memory by the application)

- if the GC could be forced to start at secure-time-known idle times

- if the application has been tested and kept as simple as possible not
to start any unhandled exception , or unpredictable situation

Would you still not trust such a RT application ?

What also could go wrong ?

Many Thanks !


-- 
"To improve is to change. To be perfect is to change often."
"You have enemies? Good. That means you've stood up for something, sometime in your life."
Winston Churchill



More information about the Mono-list mailing list