|
The errno and signal codesets Lassi Kortela (13 Dec 2022 11:38 UTC)
|
|
Re: The errno and signal codesets
Lassi Kortela
(13 Dec 2022 11:48 UTC)
|
|
Re: The errno and signal codesets
John Cowan
(14 Dec 2022 05:58 UTC)
|
|
Re: The errno and signal codesets
Lassi Kortela
(14 Dec 2022 09:08 UTC)
|
|
Re: The errno and signal codesets
John Cowan
(14 Dec 2022 22:48 UTC)
|
|
Re: The errno and signal codesets
Lassi Kortela
(15 Dec 2022 09:10 UTC)
|
|
Re: The errno and signal codesets
John Cowan
(18 Dec 2022 07:56 UTC)
|
|
Re: The errno and signal codesets
Lassi Kortela
(20 Dec 2022 10:51 UTC)
|
|
Re: The errno and signal codesets
Lassi Kortela
(20 Dec 2022 11:53 UTC)
|
The following notes apply to the Unix errno (EPERM, etc.) and signal (SIGTERM, etc.) codesets: - These facilities exist not only on Unix-like operating systems and in the POSIX standard, but also in the C language standard (https://go.scheme.org/c99) and on other operating systems that use C, e.g. Windows. I'm pretty sure they exist on VMS too, but I didn't check. Standard C mandates the existence of both the <errno.h> and <signal.h> headers, with a minimal set of standard codes in each. - I originally thought we should use the 'unix-signal symbol for the signals. But since it's standardized at the C language level and Windows does not have a native signals facility (as evidenced by Microsoft's "UNIX Application Migration Guide" at <https://learn.microsoft.com/en-us/previous-versions/ms811896(v=msdn.10)>), it's appropriate to call it by the simpler name 'signal. A third alternative would be 'posix-signal, but the codeset should not be limited to signals that have been standardized in POSIX. A fourth alternative is 'c-signal or 'csignal; IMHO this name is more confusing than helpful since people are accustomed to signals being an OS facility, not a language facility. Since the Unix signal facility has no "competitor" with the same name, there ought not to be an ambiguity. - Likewise, calling the other set 'errno instead of 'unix-errno, 'posix-errno, 'c-errno, or 'cerrno ought to be simpler for the same reasons. No language except C has a well-known facility called "errno".