Re: [PATCH v2] Test t5560: Fix test when run with dash

From: Johannes Sixt <j6t@kdbg.org>
Date: Jan 15, 2010, 19:16 UTC
Message-ID: <201001152017.00121.j6t@kdbg.org>
In-Reply-To: <7vljfzz0yd.fsf@alter.siamese.dyndns.org>

On Freitag, 15. Januar 2010, Junio C Hamano wrote:

Show 22 quoted lines

> Johannes Sixt <j.sixt@viscovery.net> writes:
> > Tarmigan Casebolt schrieb:
> >>         REQUEST_METHOD="GET" some_shell_function
> >>
> >> I can't tell from my reading of the POSIX spec whether my usage was
> >> wrong or if dash is wrong,
> >
> > According to POSIX, variables set as shown above for shell functions are
> > not exported and retain their value after the function returns.
>
> I actually looked for this yesterday, but didn't find a relevant
> definition.  But "2.9.5 Function Definition Command" [*1*] seems to
> address the issue: "When a function is executed, it shall have the
> syntax-error and variable-assignment properties described for special
> built-in utilities...".
>
> And "2.14 Special Built-in Utilities" section [*2*] says "2. Variable
> assignments specified with special built-in utilities remain in effect
> after the built-in completes...".  Taking both together, it seems that
> the assignment should be in effect after the function returns.
>
> Does my reading match yours,

These are exactly the definitions that I meant. The statement that variables are not exported is in "2.9.1 Simple Commands" http://www.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_09_01

"[If there is a command name], the variable assignments shall be exported for the execution environment of the command and shall not affect the current execution environment (except for special built-ins)."

> Yesterday, I saw rebase--interactive has a few codepaths where "output"
> shell function was used with the single-shot export; perhaps they need to
> also be fixed.

I knew these spots, and they were discussed when that code was introduced. Before I sent out the mail you were responding to, I tried various ways to show the failure in rebase--interactive, but it didn't fail...

-- Hannes