That is what I was asking about. I have a 24kb file that I was trying to
transfer in 1kb chunks, transferred 1kb without a problem but on my second call
to sftp_read there was a SSH_FX_EOF received at line 1802 in sftp.c (version
0.5.5). I built this with openssl 1.0.1. The reason I sent the ssh_session
init along with the sftp code was incase there was something I was doing wrong
in the ssh session initialization to cause this.
My read loop is as such:
char buffer[1024];
size_t length = sizeof(buffer);
size_t totalLength = 0;
size_t count = 0;
count = sftp_read(file, buffer, length);
while ( count > 0 ) {
if ( destFile.is_open() ) {
destFile.write(buffer, length);
}
totalLength += count;
count = sftp_read(file, buffer, length);
}
Kevin
-----Original Message-----
From: Aris Adamantiadis [mailto:[email protected]]
Sent: Wednesday, October 23, 2013 7:47 AM
To: [email protected]
Subject: Re: sftp_read receives EOF on second call for a 24kb file
Hi Andreas,
I think Kevin complained that sftp_read returned an EOF condition when the file
was not fully read, and if it does it's a bug in libssh.
Aris
Le 23/10/13 09:53, Andreas Schneider a écrit :
> On Tuesday 22 October 2013 18:58:17 Darren wrote:
>> Hi Kevin,
>>
>> Assuming remote to local transfer:
>>
>> You read the first chunk of data, and use sftp_seek to move the file
>> pointer
>>
>
> There is absolutely no need to call sftp_seek() you only need it if
> you resume a transfer. The API works the same way as the POSIX API.
> Files should be transferred in small chunks.
>
> #define MAX_XFER_BUF_SIZE 16384
>
> char buf[MAX_XFER_BUF_SIZE];
>
> file = sftp_open(sftp, path, O_RDONLY, 0);
>
> for (;;) {
> bytesread = sftp_read(file, buf, MAX_XFER_BUF_SIZE);
> if (bytesread == 0) {
> break; /* EOF */
> } else if (bytesread < 0) {
> /* ERROR HANDLING */
> }
>
> byteswritten = write(fd, buf, MAX_XFER_BUF_SIZE)
> if (byteswritten != bytesread) {
> /* ERROR */
> }
> }
>
> This way I can transfer files which are several gigabyte of size.
>
>
> -- andreas
>
