You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Response.close() seems to try to read all remaining data out of the socket in order to work around a flaw or limitation in the ESP32SPI socket implementation.
There are two problems with this:
A client may want / need to close an ongoing request before reading all information in it
as implemented it's not necessarily exhausting all content, but at most one chunk in the chunked-encoding
this may not be possible, the stream could be infinite
handling of chunked mode is wrong and can lead to errors during close()
As far as the last point, I did experience it 'going wrong' with the chunked mode length being part of the http stream, but there's also the case where the exact content is carefully read (runs on desktop linux):
if there's a limitation of ESP32SPI that a socket has to be fully read before it's closed, can't we push that down into ESP32SPI (or better yet fix the limitation in nina-fw) and get rid of this trouble-causing block?
The text was updated successfully, but these errors were encountered:
Response.close() seems to try to read all remaining data out of the socket in order to work around a flaw or limitation in the ESP32SPI socket implementation.
There are two problems with this:
As far as the last point, I did experience it 'going wrong' with the chunked mode length being part of the http stream, but there's also the case where the exact content is carefully read (runs on desktop linux):
if there's a limitation of ESP32SPI that a socket has to be fully read before it's closed, can't we push that down into ESP32SPI (or better yet fix the limitation in nina-fw) and get rid of this trouble-causing block?
The text was updated successfully, but these errors were encountered: