On 7 February 2012 13:45, Peter Stuge <pe...@stuge.se> wrote:
> Daniel Stenberg wrote:
>> @@ -1218,16 +1218,17 @@ static ssize_t sftp_read(LIBSSH2_SFTP_HANDLE *
>> handle, char *buffer,
>>
>> while(chunk) {
>> if(chunk->lefttosend) {
>> + if(total_read)
>> + /* since we risk getting EAGAIN below, we return here if
>> there
>> + is data available */
>> + return total_read;
>> +
>> rc = _libssh2_channel_write(channel, 0,
>> &chunk->packet[chunk->sent],
>> chunk->lefttosend);
>> if(rc < 0) {
>> - if(rc != LIBSSH2_ERROR_EAGAIN)
>> - /* error */
>> - return rc;
>> - eagain++;
>> sftp->read_state = libssh2_NB_state_sent;
>> - break;
>> + return rc;
>> }
>
> It doesn't seem like there is any point in having this loop anymore,
> since libssh2 must return after every call to _channel_write() in any
> case.
Not every call, right? Only when it returns an error.
Though I think it might be a bug setting libssh2_NB_state_sent on
every error case. Shouldn't it only happen for EAGAIN? Otherwise any
channel failure leaves expecting to continue a previous, interrupted,
read rather than starting a new one.
Alex
--
Swish - Easy SFTP for Windows Explorer (http://www.swish-sftp.org)
_______________________________________________
libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
