Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site druny.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!mtuxo!drutx!druny!weh
From: weh@druny.UUCP (HopkinsWE)
Newsgroups: net.lang.c++
Subject: Re: are inlines static?
Message-ID: <167@druny.UUCP>
Date: Mon, 7-Apr-86 11:50:48 EST
Article-I.D.: druny.167
Posted: Mon Apr  7 11:50:48 1986
Date-Received: Wed, 9-Apr-86 23:16:42 EST
References: <34200014@orstcs.UUCP>
Organization: AT&T Information Systems Labs, Denver, Co
Lines: 36
Summary: names in file scope are extern by default

In article <34200014@orstcs.UUCP>, nathan@orstcs.UUCP (nathan) writes:
> static inlines:
> 
> alice!ark claims that "inline" routines are not implicitly
> static, unless declared otherwise.  He/she may be right, in
> the case of the existing compiler, but Rich Wales is right "by the
> book".
> 
> Read section r8.1 more closely: *everything* at the
> top level is static unless declared otherwise.

Section r8.1 is talking about storage class only, not name visibility.
The pertinent sections are r4.1, which defines three kinds of scopes:
local, file, and class. File scope is that for all objects declared
outside of a block or class. It is visible after the point of declaration.
Section r4.3 on linkage states that "A name of file scope that is not
explicitly declared static is common to every file in a multi-file
program; so is the name of a function. Such names are said to be external."

> Doesn't the
> existing compiler follow this?

Mine behaves exactly as specified above (currently that distributed
outside AT&T).

Regarding inlines specifically, section r4.3 gives more details,
but the nature of inlines are that they must be defined in each
file where they are used, else the compiler couldn't possibly
perform inline expansion.

			Bill Hopkins rm. 30G11
			AT&T Information Systems Labs
			11900 N. Pecos St.
			Denver, CO 80234
			(303)538-4944
			ihnp4!druny!weh