taos_fetch_block _v3 is slow

[hadrianl]

https://github.com/taosdata/taos-connector-python/blob/4ab57be79615accd9bdc5868fe361a87b91642c1/taos/cinterface.py#L504

I just runned some test and found that taos_fetch_block_v3 is pretty slow here, especaily :

• _crow_binary_to_python_block_v3
• _crow_nchar_to_python_block_v3
• taos_is_null

it seems too much cast, too much python loop.it cost at least 80% time to parse the data from c to python. is it any plan to improve this?

[sangshuduo]

Would you like to contribute to improving it? :)

[hadrianl]

I would try to submit a PR when I`m available

[hadrianl]

is it ok to accept the code like this to improve the performance:

• Cython ,setuptools and compiler are needed
• Maybe make it optional when fetch data into python object.
• In my benchmark test, 2~4 times faster than using ctypes

[markcheney1]

is it ok to accept the code like this to improve the performance:

• Cython ,setuptools and compiler are needed
• Maybe make it optional when fetch data into python object.
• In my benchmark test, 2~4 times faster than using ctypes

用了pr合入的代码，速度有至少100%的提升，非常了不起。目前我也在探索数据读取的加速，要不然用taospy做回测就要非常久