Problem when retrieving huge amount of response.

[plurkey]

It works great when a plurk only has only a few responses (roughly less than 100).

But as the amount of responses grew, memory got drained dramatically. With a bit testing, I found out that it needs about 1G of memory to be able to retrieve 500 responses successfully and than call delegate method - (void)plurk:(ObjectivePlurk *)plurk didRetrieveResponses:(NSDictionary *)result.

Is there a way to solve this issue? Thanks in advance.

[zonble]

Actually I need some test cases to know about the memory usage while there is an extraordinary amount of incoming responses.

Maybe we can try not to fetch all the responses in the first time but to set a limit in every HTTP request. I just noticed that the /response/get API has a parameter, from_response, which allows us to set a offset. So, if there are 500 responses, we may fetch the responses with 5 requests, and each request fetch 100 responses. I guess it may help.

[zonble]

Oh, wait. Plurk API only allows us to set offset, but we are unable to set limits. Maybe what we can do is to make another autorelease pool to release unused object more quickly while parsing the fetched data.

[plurkey]

Does this means that the only way to solve this issue is to wait for the author of bsjsonadditions to update ?

I know account that often have 500+ response plurks you can try out.
http://www.plurk.com/plurktaiwan

[zonble]

Maybe we can fork bsjsonadditions, but I am not sure how much does it help.

[plurkey]

I tried your suggestion and added an auto release pool in - (BOOL)scanJSONObject:(NSDictionary **)dictionary method. That completely solves the burst memory usage issue, but the speed performance seems to lower a bit.

[plurkey]

I've forked bsjsonadditions and added the modification mentioned above, anyone that is interested may refer to my repo.
http://github.com/plurkey/bsjsonadditions