Are you using Vista 
?
With that Memory load i will never ever touch it.
Hahaha, no
Vista doesn't run with that little memory. Assume at least twice as much for Vista.
Those are figures from my XP system, instrumented display driver, 2 extra drivers for 6DOF HIDs, Tortoise, and 2-3 services beyond the bare minumum.
python, it is very much like C
One major thing that is not like C is using indent for delimiting blocks and scopes. Some people even see this as an advantage (the Python devs explicitely designed it this way), but to me it is a showstopper which prevents me from considering Python for anything "serious". Maybe I'm wrong, but I'm getting the creeps if whitespace is used as control structure
Another thing is the entire bytecode story. Python is bytecode ("compressed ascii" as you said). Inevitably, bytecode/interpreted is slower (on the order of 5-15 times) than native code.
Yes, the bytecode advocates tell you that this does not matter because CPUs get twice as fast every 18 months. Sun has been saying this every 2 days since they released Java in 1995. Without any doubt, they sold millions of CPUs due to Java, too. Let's be honest, wasn't this the purpose of Java in the first place?
But let's face it, Java performance still sucks on 2007 CPUs. The same is true for C# and all the other bytecode languages. This isn't surprising, either. Demand grows faster than hardware. Today's programs and operating systems do a lot more (and often useless) stuff than they did 10 years ago. CPU cycles drain away, and the time when the linear computing power steadily increased is over, too. Quad core CPUs don't help if your single-threaded bytecode program is just too slow.
If I launch Eclipse on my notebook, it takes about two minutes to start. I have no flipping idea what the hell is taking so long just to bring up the IDE, but maybe it's thrashing virtual memory (the notebook has only 1 GB) or something? Code::Blocks takes 5 seconds to start on the same system (and this is already too much).
Or look at nVidia's FX Composer. Just compare version 1.8 (a really great tool) to version 2.0 (now written in C#): 31 MB / 2s startup versus 103 MB / 26s startup. Don't even ask how long it takes to load a workspace. Sorry, but I simply can't take a program for serious, if I can go to the kitchen and get a fresh cup of coffee while it loads.
Hmm... Ok, that's what it was like in 1985, and sometimes you had to switch floppy disks in between too... but hey, we're not in 1985 any more.
if Boa with all its functionalities is just 38 Mb and CodeBlocks is 35 this should mean something about python vs C
It means nothing, really. Code::Blocks rigorously over-allocates memory (in figures of 1000s). Boa/Python might do just the same thing, or it might do something else. In any case, 35 and 38 are on the same order of magnitude. If one of them took 20 MB and the other took 200 MB, then maybe you could tell something, but even then it's hard, since it's not a good measure.
Don't get me wrong, if you think that Python is the thing for your project, and you feel comfortable with it, then go for it. It's better to use something one is comfortable with than something one doesn't feel right with.
I'm just saying that it would not be
my personal choice. I think Python is ok for quickly hacking together a Blender plugin, or for scanning a text file from a
cron job (great for people like me who find Perl's syntax repulsive). Just for a "real" application.... I'm not so sure.
