GSOC Enquiry

[Gedare Bloom]

On Fri, Mar 16, 2012 at 5:09 PM, Joel Sherrill
<joel.sherrill at oarcorp.com> wrote:
> Hmm.. how did you try to email me?
>
> Best to discuss this on the list anyway.
>
> Basically the pthread keys need to be hashed based on something unique like their id or TCB pointer. The reason is that each task needs a key slot and when in unlimited task/thread allocation mode, the key table is not resized.
>
> The hash code needs to be in the supercore and also promoted via a proper user api with docs. The algorithm in the hash needs to be low algorithmic order for lookup. Insertion and deletion can incur work but modest is desired.
>
> Gedare.. is an rbtree useful at all here?
>
I haven't looked too closely at this problem/project. I would have to
get a better idea of the usage. The rbtree would support a dynamic
map, but the lookup/insert/delete all would be o(log n). Hashing would
only give a good way to derive search keys.

> --joel
>
> Aaron Kalair <aaronkalair at gmail.com> wrote:
>
>>Hi,
>>
>>I'm interesting in applying to RTEMS as a Google Summer of Code 2012 student.
>>
>>I've taken a look at the list of open projects and I like the look of this one:
>>
>>http://www.rtems.org/wiki/index.php/UseHashOrMapInNotepadsAndKeys
>>
>>Would someone be able to discuss this project in more detail with me? I have tried to email Joel Sherrill but I got a 550 550 5.1.1 User unknown error.
>>
>>How difficult do you think this project would be to complete? I have experience with coding in C and I'm familiar with the hashmap data structure but not classic API notepads and POSIX keys.
>>
>>
>>Thank You
>>
>>Aaron Kalair
>
> _______________________________________________
> rtems-devel mailing list
> rtems-devel at rtems.org
> http://www.rtems.org/mailman/listinfo/rtems-devel