From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 29280 invoked by alias); 5 Jul 2002 03:34:27 -0000 Mailing-List: contact gdb-patches-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sources.redhat.com Received: (qmail 29272 invoked from network); 5 Jul 2002 03:34:26 -0000 Received: from unknown (HELO www.dberlin.org) (138.88.47.74) by sources.redhat.com with SMTP; 5 Jul 2002 03:34:26 -0000 Received: by www.dberlin.org (Postfix, from userid 503) id D0A131847991; Thu, 4 Jul 2002 23:34:22 -0400 (EDT) Received: from localhost (localhost.localdomain [127.0.0.1]) by www.dberlin.org (Postfix) with ESMTP id 41C241837801; Thu, 4 Jul 2002 23:34:14 -0400 (EDT) Date: Fri, 05 Jul 2002 03:39:00 -0000 From: Daniel Berlin To: Jim Blandy Cc: Petr Sorfa , "gdb-patches@sources.redhat.com" Subject: Re: [PATCH] Support for multiple DWARF comp unit headers In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Spam-Status: No, hits=-1.7 required=5.0 tests=IN_REP_TO,NO_MX_FOR_FROM,AWL version=2.31 X-Spam-Level: X-SW-Source: 2002-07/txt/msg00098.txt.bz2 On 4 Jul 2002, Jim Blandy wrote: > > Just out of curiosity --- after we've built `struct die_info' objects > for a CU, we don't need the abbrev table any more, do we? I may be > misremembering the way the code behaves, but it seems to me that one > could construct an abbrev table before calling read_comp_unit, and > then throw it away when done. Perhaps the abbrev tables are not big > enough for this to be worth thinking about. They generally aren't. Two factors: 1. Generation of abbrev tables is usually very optimized. 2. The structure of debug info is fairly regular, it's only the actual information present in the structure that changes. THus, the abbrev tables are fairly small. In fact, i've never seen one that was more than about 150 entries.