Python is just as good as C++ for real apps

[Chris Liechti]

"Thomas Heller" <thomas.heller at ion-tof.com> wrote in
news:a36v9a$169s93$1 at ID-59885.news.dfncis.de: 
> "Chris Liechti" <cliechti at gmx.net> wrote in message
> news:Xns91A5D295BC720cliechtigmxnet at 62.2.16.82... 
>> "Thomas Heller" <thomas.heller at ion-tof.com> wrote in
>> news:a36j1b$15eigm$1 at ID-59885.news.dfncis.de: 
>> > What would you think about python plus raw pointers included?
>>
>> crashes? direct access to memory can be practical but is very error
>> prone... 
>>
> 
> Of course.
> 
> But, good as Mark Hammonds win32 extensions are, you can also crash
> Python easily with them. Maybe because it's impossible (even in the 
C
> layer) to validate all the arguments you are passing to a function
> call, maybe because it's just windows...

:-)
 
[...]

> Sure, it's not complicated in simple cases, you can (even now) do it
> with the struct module (which also knows about the C compiler 
alignment
> rules).
> 
> It gets tougher when you have nested structures, structures 
containing
> pointers, and so on.

well i had structures of structures and arrays of structs etc...
i can write single values or entire structs/arrays as a list of bytes.
using it to modify the memory is no problem. my first version was 
python only. it parsed the .h and created a nested class hierarchy. 
however, because of the alingnment thing, i decided to use the c 
compliler and sizeof/offsetof macros to get the correct information.

i think one of the bigger problems is parsing c code (header files) to 
extract the struct (defines, typedefs, ...)

let us know if you have something usable. i can also show you my 
python wrappers if you're interested (the .h parser is not that 
reliable)

chris
-- 
Chris <cliechti at gmx.net>