On Fri, Jan 27, 2017 at 04:24:47PM +0300, Dan Carpenter wrote:
On Fri, Jan 27, 2017 at 02:08:06PM +0100, Johan Hovold wrote:
On Fri, Jan 27, 2017 at 03:47:21PM +0300, Dan Carpenter wrote:
The name sucks is what I'm saying. Please fix it eventually.
I disagree. The name uses a common prefix that reflects the object it is working on. This should not be changed. As it is currently named, the function is also self-documenting. Trying to abbreviate the name just to meet a 80 col limit (when there are alternatives) is just silly.
gb_interface_request_mode_switch()
It's not self documenting because there are no verbs in that sentence.
"request" is a verb.
We're requesting that the remote interface initiates a mode switch (something which it can refuse).
It's a function that is only a called a couple of times. No need to rewrite it as gb_intf_rms() or such (even if gb_intf would be a reasonable prefix).
You comments about missing verbs etc make me wonder how you parse the name:
gb_interface - a common prefix, indicating object being acted on request_mode_switch - describes what is done
we're not fetching anything as your "gb_get_mode_switch()" suggestion indicates you read it as.
I guess in JAVA it would be considered good style.
This is "Enterprise Quality" in the derogatory sense of the term. The prefix makes everything too long and doesn't add any value. We don't keep the 80 character limit because of small screens, we keep it to discourage this type of code.
And to keep indentation down.
The kernel already has similar function names such as "usb_driver_set_configuration" or "pm_runtime_mark_last_busy" which follow the same pattern (prefix - verb - object).
Johan