acl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Acl-devel] [PATCH 0/2] Handle quoting of backslashes more clearly


From: Andreas Grünbacher
Subject: Re: [Acl-devel] [PATCH 0/2] Handle quoting of backslashes more clearly
Date: Sat, 5 Dec 2015 23:45:55 +0100

Jeff,

2015-12-02 16:52 GMT+01:00  <address@hidden>:
> From: Jeff Mahoney <address@hidden>
>
> We recently had a bug report where a customer required the use of numeric
> usernames as part of a Windows domain.  The usernames would take the form
> of DOMAIN\123456, which resulted in the acl unquoting code interpreting
> them as DOMAINS456.  The customer ultimately discovered that they could
> work around the issue by escaping it manually by using DOMAIN\134123456.
> They must've discovered it by reading the source for the project since
> it's not documented in the man page.  I think we can agree that it's not
> a great user experience.
>
> The following two patches try to improve that user experience by allowing
> the escaping of literal backslashes in usernames (i.e. \\).

I'm fine with that, thanks for the fix.

> I've also
> added test infrastructure to handle testing for usernames like this by
> adding a helper library that provides the getpw* and getgr* routines
> used in the test code and redirects the reads to project-local copies of
> passwd and group.  I'll admit it's a bit overkill, but while it worked
> well to choose standard system usernames for testing like 'bin' and
> 'daemon', I needed a way to test the quoting code and that required
> non-standard usernames.

Did you write this just for libacl? Would it make sense to move it
into a separate package to allow other test suites to use it as well?
For example, xfstests depens on a local user / group "fsgqa".

Other than that, the code looks okay, so I'll add your two patches in
their current form for now.

Thanks,
Andreas



reply via email to

[Prev in Thread] Current Thread [Next in Thread]