Close in uartdmx is checked for return value...

[rewolff]

Close is usually not checked for return value, as there is little you can do if close fails.

That said, the current code only tries to issue a warning if such happens. However... it won't work (if I am correct in reading the code).

https://github.com/OpenLightingProject/ola/blob/master/plugins/uartdmx/UartWidget.cpp#L86

Close is documented to return -1 on error, and zero on success. That means > 0 will never happen.

Maybe != 0 was intended. I personally like to check for < 0, allowing for calls to return error status-es other than -1, but for over 30 years now, most system functions are documented to return -1 on error and nothing else.... So my personal preference:
if (close(m_fd) < 0) {
probably more consistent with other code:
if (close(m_fd) != 0) {
otherwise valid:
if (close(m_fd) == -1) {

[peternewman]

It looks like we've gone for a mix of:
if (close(m_fd)) {
and

int ret = close(m_fd);
if (ret == 0) {

and

int ret = close(m_fd);
if (ret != 0) {

In other code.

Do you want to open a Pull Request and apply the fix @rewolff ?

[peternewman]

@rewolff are you willing to put in the fix for this then?

[rewolff]

Not now.
Keep reminding me. One of these days I'll get fed up and do it. Ok?
(I sell DMX boards and we've sold a batch that ALL don't work. Recall or try to fix on customer premises.... )

[peternewman]

Okay @rewolff . Thanks for thinking about it. :) We've got 0.10.3 coming out imminently if you fancied getting it into that. :p

Something major on the boards, or just a missing link?

[rewolff]

I'm going for
if (close (...) )
The reason being that it is the shortest. It is a common idiom, and if you have to learn that idiom, you need to learn it only once throughout the whole code if we apply it constently.