On 02/03/2016 09:55 AM, Michael Kerrisk (man-pages) wrote: > Hi Carlos, > > On 01/27/2016 08:16 PM, Carlos O'Donell wrote: >> The goal is to make it easier for a reader to follow along with the >> documentation, prototypes, and examples that use "struct sockaddr". >> We achieve this by using a consistent naming for "struct sockaddr" >> -based types. Instead of using "sa", or "name" or "local" we use >> "saddr", "name_saddr" and "local_saddr". We avoid variable names >> like "sa" which are used for "sigaction" examples. >> >> Instead of the generic "addr" we use "saddr" in places where we accept >> "struct sockaddr" as an argument, either the struct or a pointer to the >> struct. Where the eventual goal is to cast the variable to a >> "struct sockaddr"-based type, such variable names are adjusted. So for >> example if we have a "struct sockaddr_un" variable we call it >> "local_saddr" if it will eventually be cast to "struct sockaddr *". >> >> We might have standardized on just "addr" but that's ambiguous >> and I'd like to use, where appropriate, slightly different variable >> names for the various forms of "struct sockaddr" like "sockaddr_un", >> "sockaddr_in", "sockaddr_storage" and others, so "addr" is a poor >> choice when helping the reader follow along (not to mention the >> confusion with virtual memory addresses and mmap). >> >> Please apply. >> >> Patch against master. > > I'm not quite convinced about this patch. Some thoughts: > > * Consistency is a good thing. Names such as 'a' or 'sa' are > not very helpful. And I'm happy to see that stuff go away. Agreed. > * I'm reluctant about the odd name 'saddr'. It's not consistent with > the BSDs, or much existing documentation. Also, I'm not convinced > that the possible confusion with VM addresses or mmap().) So, if > standardizing, I'd prefer to stick with 'addr' (which is also > consistent with 'addrlen'). > > I applied the patch below. Perhaps that's enough change to make > you happy. I'm not dead set against renaming 'addr' and so > forth, but the argument for the benefit would need to be fairly > convincing. Your patch looks great. Any consistency is a step forward. Regarding 'addr' vs 'saddr', your argument for consistency with BSD is the strongest one I've heard so I'm happy to see us use 'addr.' Cheers, Carlos. -- To unsubscribe from this list: send the line "unsubscribe linux-man" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
