installation

It could be argued that the real root of the problem is not using Harvard architecture designs with separate instruction and data spaces. But there's no way back the only way is forwards and that means coping with the past as it happened.

Indeed, and the same thought was forming in my brain as you were probably writing that.

However there are advantages to being able to mix code and data.

I am not sure if a basic multitasking scheduler could be built without the ability to use data to modify code as it were. Still its early, so I could well be wrong there.

This is of course true. But the very idea of 'have to remember' when a compiler can do that for you is icky - in my point of view.

To get back on the pi-track, I have written several things in python, from utilities to small bot-systems betting on UK/IE horse races, but the lack of compiler makes me do mistakes (like indentation, or trying to print an int on stdout without conversion) which then becomes run-time crashes.

Yes I do know about pylint, but it has to be configured and set up, which I don't HAVE to do, thus I don't do it.

A strict compiler, that is my best friend.

Clever. But really. Should one have to resort to this ? Strict compilers for the people, please.

I do agree. That picks up most of the inadvertent casting issues.

But it doesn't pick up buffer overflows.

Over the years I have basically asked myself two questions

(a) Is it possible the data going in this buffer could be larger than the buffer? (b) If so, what would happen if I restricted the data to the buffer size, or threw an error?

There are other options,. like always using malloced data after doing a len() or strdup or whatever, but these all imply that you already have the data in a suitably large buffer..

If its coming in a byte at a time, you have to accept the overhead of checking it its out of bounds ...but that is better than having run time checking enforced by the language whether you need it or not.

If the compiler knows enough, that is if you have a strongly types language, the effect will actually be that some run time checks are eliminated.

a simple for loop in c over an array

int len = 200; int data[len] ;

for (int i=0; i

I have looked at the download of Rasbian and nowhere can I see what version it is. How do I know it will be Jessie?. I have a new FAT32 formated micro SD card which I can use. As I remember the card I have in use at the moment was bought preloaded when I bought the Pi2 and has nothing of value on it except the WiFi and printers are set up.

Malcolm

In article (Dans l'article) ,

For French speaking people we have the 2016-02-09-raspbian.zip available. It's noted to be Jessie "2016-02-09-raspbian-jessie.img"

Thanks Jean-Pierre. Currently updating but stuck at the Wolfram Engine -waiting for headers as others have mentioned. It is jessie that is being down-loaded but not sure what version Malcolm

And the return address is data, so it's still vulnerable. Most programs can be induced to commit mischief by suitably altering the return address.

True. Security bugs are the result of sloppy programming. C just makes sloppiness easy. ;-)

Think of C as a gun without a safety...convenient but dangerous in undisciplined hands.

the current noobs installs Jessie & is probably the easiest option

I shut down then rebooted and everything zooming along. When it was stalled the green 'activity LED' was giving a very slow weak pulse but its usual bright self at present

Malcolm

It. Is. Right. There. WTF.

Some of the other download sites perhaps are not as clear? Anyway, even with the specified site, it's usually good practice to install updates, etc via app-get.

Yes, I did find this after my post.

In article (Dans l'article) ,

I find now a Web page explaining all what you want to know but didn't dare ask...