Re: [gnet-dev] gnet_io_channel_writen() clarification
- From: "David A. Helder" <dhelder gizmolabs org>
- To: Scott Long <scott swiftview com>
- Cc: gnet-dev lists gnetlibrary org
- Subject: Re: [gnet-dev] gnet_io_channel_writen() clarification
- Date: Tue, 09 Dec 2003 20:30:11 -0500
I've finally investigated this. This was delibrate. If error is
G_IO_ERROR_NONE and bytes_read/written != n, then EOF was received. The
programmer should check this.
David
On Thu, 2003-10-30 at 17:50, Scott Long wrote:
> However I've encountered a small inconsistency in the documentation
> for gnet_io_channel_writen().
>
> The comments in the source and the online documentation both say: "If
> bytes_writtenp is set, the number of bytes written is stored in the
> integer it points to."
>
> To me, this implies that if bytes_writtenp is NULL (i.e. "not set"),
> then the number of bytes written will not be stored.
>
> However, in the source for this function (iochannel.c) there is the
> test:
>
> g_return_val_if_fail (bytes_writtenp, G_IO_ERROR_INVAL);
>
> Thus this function cannot be used if bytes_writtenp == NULL, i.e., if
> the caller doesn't care about the number of bytes written. This is
> perfectly fine behavior; but the documentation and reality are not in
> sync. If it is required that bytes_writtenp point somewhere valid,
> then the perhaps the documentation should be updated to make it
> clearer.
>
> A similar inconsistency exists for the other function,
> gnet_io_channel_readn().
--
David Helder - dhelder gizmolabs org - http://www.gizmolabs.org/~dhelder
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]