Re: [PATCH] Add seuser and user_extra to package format

[Ivan Gyurdiev <ivg2@cornell.edu>]

>> I will resync my nodecon patch tomorrow - let me know if anything's 
>> wrong with it, otherwise will submit as is (ordering pending in a later 
>> patch).
>>     
>
> I still wasn't clear on the disposition of the pylist patch.
>   
I think the typemap code would be sufficiently different in diff. 
languages, that we can (and will have to) write a completely new .i file 
for another language. I doubt the current typemap code is language 
portable (I don't know how adding things to output vector will translate 
to other languages). It seems to me that the portable portion of the 
code is generated by swig, and any typemaps we add on top will be mostly 
language specific.

Dan, maybe you can comment on this. I added a patch that translates 
out-bound C list (type**, int) to an actual PyList, which is garbage 
collected, and you can have a for-in loop range over it without an 
index, and you don't need get_*_by_idx. There are probably other 
advantages that I don't know about. Would this be useful to a python 
developer? What about an in-bound PyList (for set_roles)?  Do iterate() 
function pointers work currently? If not, maybe better python 
integration can help make them work?

> For the nodecon patch, I wasn't sure if we were sufficiently confident
> in the ABI/API stability to include it for FC5 or if it should wait on
> FC6 devel.  It would likely help to have feedback from potential users
> of the nodecon API.
>   
Ok... I would like to hear any recommendations on byte order/byte array 
handling.

--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.