Dmitriy

Many differences between python and cython implementations. ``` aiokafka/record/_crecords/legacy_records.pyi:38: error: Class aiokafka.record._crecords.legacy_records.LegacyRecordBatch has abstract attributes "compression_type", "timestamp_type" [misc] aiokafka/record/_crecords/legacy_records.pyi:38: note: If it is meant to be abstract, add 'abc.ABCMeta' as...

https://github.com/aio-libs/aiokafka/blob/master/aiokafka/record/_crecords/default_records.pyx#L452 https://github.com/aio-libs/aiokafka/blob/master/aiokafka/record/_crecords/default_records.pyx#L459 differences from python

@ods please reopen this issue :)

4. I think, the main problem is not result of as_datetime/as_date/as_time. Make sense to refactor internal representation of datetime (DataType::*) for solving problem 1904. And if somebody adds this flag,...

> > 4. Make sense to refactor internal representation of datetime (DataType::*) for solving problem 1904. > > Could you explain a bit more what you mean here. For context...

> This comes at the cost of memory, however, as all the data would be in memory. This could prove an issue for very large datasets and causing OOM errors....

Xls/Ods don't support lazy iteration, see https://github.com/tafia/calamine/pull/370#issuecomment-1789074220

I saved example file using LibreOffice with deutsche interface - [errors_de.ods](https://github.com/tafia/calamine/files/12583086/errors_de.ods). There is part of `content.xml` of this file below: ```xml #DIV/0! #NV #NAME? Fehler:509 #ZAHL! Fehler:502 #WERT! ``` and...

May be we should check cells.len() == isize::MAX in loop, when add cell to `cells`? And this should be backported to other formats too.

@SgtPepper47, sorry, cells.len() == usize::MAX