On 16/03/10 07:55, Steven Dake wrote:
> On Tue, 2010-03-16 at 18:53 +1100, Angus Salkeld wrote:
>> On Tue, 2010-03-16 at 00:34 -0700, Steven Dake wrote:
>>> On Tue, 2010-03-16 at 18:30 +1100, Angus Salkeld wrote:
>>>> On Tue, 2010-03-16 at 00:14 -0700, Steven Dake wrote:
>>>>> On Tue, 2010-03-16 at 17:47 +1100, Angus Salkeld wrote:
>>>>>> Hi
>>>>>>
>>>>>> I found this a handy library whilst writing the cpg_test_agent
>>>>>> for the test harness. Others might find it useful too.
>>>>>> We currently have quite a few exposed (generic) libraries
>>>>>> perhaps this could be one of them (logsys, ipc/s, etc ...).
>>>>>>
>>>>>
>>>>> coropoll is exported currently via a header file, although I am not
>>>> sure
>>>>> it is packaged as a DSO.  I had hoped people would start to use the
>>>>> non-cluster APIs available in Corosync for their own client/server
>>>>> applications, but as a developer when I think Corosync I think
>>>> "cluster
>>>>> software".  This thought pattern may be a general condition of the
>>>> other
>>>>> client/server-ish apis in Corosync.  It may be that providing those
>>>>> reusable components is outside the mission and general audience of
>>>> the
>>>>> Corosync community.
>>>>>
>>>> It might be more useful to remove these more general parts into a
>>>> separate library like many other projects have done. It will
>>>> help to harden the library and focus corosync on it's prime
>>>> mission (clustering).
>>>>
>>>> -Angus
>>>
>>> They already are separate libraries (except coropoll).  The big things
>>> missing are docs, examples, test cases, etc.  ATM I'd rather spend my
>>> efforts on fixing up the documentation for corosync, providing test
>>> cases there, and examples.
>>
>> Maybe I was not clear, move these libraries into a separate project.
>>
>> This will help in people adopting them. At the moment you have to
>> install corosync if you want to use some cool client/server libraries.
>> Unless you are using corosync this really does not make sense.
>>
>> Also this could let someone else worry about the examples, test cases,
>> docs for the library :)
>>
>
> interesting idea, volunteering to maintain it?
>


I think that's a good idea. there are a lot of programs with their own 
socket polling code (I must have written it 4 or 5 times myself!), so a 
well-tested library that does all the right things would be a boon I 
think. Not that I necessarily have the time to maintain it :S

Chrissie
_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to