|
From: | Paul Eggert |
Subject: | Re: master fails to build on FreeBSD when ACL support is on |
Date: | Mon, 22 Jan 2018 09:02:47 -0800 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.5.2 |
On 01/22/2018 07:52 AM, Eli Zaretskii wrote:
I didn't mean to change the API, I meant to ask why doesn't copy-file use acl_errno_valid, and if that says the error just means ACLs aren't supported in this case, silently gives up on copying ACLs? The return value doesn't need to change. We use set-file-acl when we need a more fancy copying, which we do in Lisp, so why not in the primitive?
You asked why the two functions are inconsistent, and my answer was that the inconsistency springs from the fact that they have different APIs. The two functions would continue to be inconsistent even if we were to change the behavior in the way that you suggest, since set-file-acl would inform callers whether ACL setting failed (by returning nil instead of t), whereas copy-file would not. If consistency is the goal then we need to change their APIs somehow.
[Prev in Thread] | Current Thread | [Next in Thread] |