Luffbee

> No, I want a document describe the **format string** of `xqc_log`. For example, when I want to print a `uint32_t`, which format specifier should I use? I used `%u`...

This link is helpful, but I think it's still missing some details, for example, how should I print `uint64_t`? I looked [`src/core/ngx_string.c`](https://github.com/nginx/nginx/blob/f7ea8c76b55f730daa3b63f5511feb564b44d901/src/core/ngx_string.c#L90-L119) (mentioned in the link), it gives a better...

> Want to ask if you have solved the problem?I'm having the same problem. I solve it by using branch 8.3.1 of Tongsuo.

I'm not familiar with the internal buffers in asyncssh. For the buffer of pipewriter in this case, I think it is not implemented by asyncssh, and I didn't find a...

Thanks for your repair. I will try it in the next few days. However, I think the fix that supports recv_eof will cause some "output synchronization" problems: writes on the...

I tried your repair and it works for the case I described at the beginning. However, I found another problem caused by the `connect_write_pipe/connect_read_pipe`: the side effect of them is...

I agree that saving and restoring the non-blocking state won't work. However, I still think blocking write on `sys.stdout` is ok, as it usually won't block too long. Like you...

> The problem with sys.stdout is that the system WON'T block on it. I didn't make it clear. My suggestion is never to use `loop.connect_read_pipe()` or `loop.connect_write_pipe()` on any FD,...

> Hi! I'm not reproducing this in `plotly==6.1.2`, could you try this out again with that version? I cannot reproduce the problem either. I think it has been fixed. I...