feat(binding/c): Add blocking_reader for C binding

[jiaoew1991]

issue: #3257

[xyjixyjixyji]

To be more specific, we could have two functions to reach this approach

1. opendal_reader_new() to construct a new Reader
2. opendal_reader_buffer_read() to read through a Reader

For the first function, you could refer to the opendal_operator_new(). For the second part, your implementation in this PR will not change much.

WDYT?

[Xuanwo]

2. opendal_reader_buffer_read() to read through a Reader

Negative. We should only have opendal_reader_read() that accepts an address passed by the C side.

[xyjixyjixyji]

2. opendal_reader_buffer_read() to read through a Reader

Negative. We should only have opendal_reader_read() that accepts an address passed by the C side.

Hmm, so we setup the reader (buffer size, path, etc.) on the first function, and then do the read that follows the configuration?

[Xuanwo]

Hmm, so we setup the reader (buffer size, path, etc.) on the first function, and then do the read that follows the configuration?

The input write buffer size should be passed in every read call the same as in this PR.

I am concerned about the API semantics. We will not offer opendal_reader_buffer_read() because it implies that we will have a opendal_reader_read().

[jiaoew1991]

Hi @Xuanwo @Ji-Xinyou , should I improve this pull request, or should I open a new one? 🤔

[Xuanwo]

should I improve this pull request, or should I open a new one? 🤔

Both LGTM, I think you can work on this PR directly.

[Xuanwo]

Others LGTM, cc @Ji-Xinyou would you like to take a review?

[xyjixyjixyji]

All other parts LGTM, except for the return value of reader read. I mean it is always good to have as much error information as possible 🤣

[Xuanwo]

Hi, @jiaoew1991, #3250 is about to merge. Maybe you can work upon it directly?

[jiaoew1991]

@Xuanwo Updated it, but most of the CI jobs are cancelled 😅

[xyjixyjixyji]

I closed and reopened this PR to rerun the CI.

[Xuanwo]

I found an issue that our naming for structs and functions is inconsistent. The word blocking shown up randomly.

cc @Ji-Xinyou, how about removing all blocking from our API for now? We can add async version in the future.

[xyjixyjixyji]

I found an issue that our naming for structs and functions is inconsistent. The word blocking shown up randomly.

cc @Ji-Xinyou, how about removing all blocking from our API for now? We can add async version in the future.

Sure, that could be the default case since blocking is the most common.

When we add async in the future, we can add async keyword in the API.

[Xuanwo]

Sure, that could be the default case since blocking is the most common.

When we add async in the future, we can add async keyword in the API.

Cool, would you like to submit a PR after this one merged?

[Xuanwo]

Thanks