From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 8862 invoked from network); 26 Oct 1999 13:35:52 -0000 Received: from sunsite.auc.dk (130.225.51.30) by ns1.primenet.com.au with SMTP; 26 Oct 1999 13:35:52 -0000 Received: (qmail 21165 invoked by alias); 26 Oct 1999 13:35:39 -0000 Mailing-List: contact zsh-workers-help@sunsite.auc.dk; run by ezmlm Precedence: bulk X-No-Archive: yes X-Seq: 8427 Received: (qmail 21142 invoked from network); 26 Oct 1999 13:35:38 -0000 Subject: Re: PATCH: completion In-Reply-To: <3815A627.19C7F2A@u.genie.co.uk> from Oliver Kiddle at "Oct 26, 1999 2: 1:27 pm" To: opk@u.genie.co.uk (Oliver Kiddle) Date: Tue, 26 Oct 1999 14:35:14 +0100 (BST) Cc: zsh-workers@sunsite.auc.dk X-Mailer: ELM [version 2.4ME+ PL48 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Message-Id: From: Zefram Oliver Kiddle wrote: >Note that all the .o files have two dots (e.g. modentry..o) when >building on >AIX. This is probably a mistake somewhere but it has always compiled >happily so >I've never bothered. This should probably be documented in the development guide or somewhere. In the very first dynamic linking patches, all objects were ".o", as usual. However, objects need to be compiled with different options when they're in a module than from when they're linked into the main executable. After changing the list of compiled-in modules, it was necessary to remove all the objects. So we changed it so that objects to be linked into modules got the suffix "..o", so that both lots of object files could coexist, and even when you don't need both lots to exist at once it avoids a lot of potential confusion. The module install directory was changed from $prefix/lib/zsh to $prefix/lib/zsh/$ZSH_VERSION at the same time, for similar confusion-avoidance reasons (plus coexistence of multiple installed zsh versions). Ever since then, I've been glad that we got that right from the start. I hate to think, if we hadn't done these two things, how many support queries we'd be answering where it turned out that people had got the wrong objects linked together. (This is all related to the principle of least surprise; I'm going to dub it the "principle of least support".) -zefram