Mailing-List: contact cygwin-apps-help AT cygwin DOT com; run by ezmlm Sender: cygwin-apps-owner AT cygwin DOT com List-Subscribe: List-Archive: List-Post: List-Help: , Delivered-To: mailing list cygwin-apps AT cygwin DOT com To: Robert Collins Subject: Re: setup w/char* eliminated is big Message-ID: <1013985230.3c702fce8c11b@mail.syntrex.com> Date: Sun, 17 Feb 2002 23:33:50 +0100 (CET) From: ptsekov AT syntrex DOT com Cc: "Gary R. Van Sickle" , cygwin-apps AT sources DOT redhat DOT com References: <013801c1b6e8$6eb8d780$0200a8c0 AT lifelesswks> In-Reply-To: <013801c1b6e8$6eb8d780$0200a8c0@lifelesswks> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit User-Agent: IMP/PHP IMAP webmail program 2.2.7 X-Originating-IP: 212.171.97.58 Well, after looking at the code for some time and playing with it i finally discovered that the purpose for the 100kb increment is not the String class but this: This is taken from the Changelog for v.2.47 of the Makefile.in Backout -fno-exceptions option. This explains it all! No need to worry ;)) Bad thing is I havent looked at the Makefile.in first, but spent time looking at assembly shite ;))) Quoting Robert Collins : > Well, 11K (i.e. 2 seconds download at 56Kbit) is neither here nor > there. > 100K is more of an issue. > > Anyway, Chris' response (AFAICT) implied that the size was not an > issue, > but cross-compilability was. > > Rob > > > === > ----- Original Message ----- > From: "Gary R. Van Sickle" > To: > Sent: Saturday, February 16, 2002 6:20 PM > Subject: RE: setup w/char* eliminated is big > > > > > -----Original Message----- > > > From: cygwin-apps-owner AT cygwin DOT com > > > [mailto:cygwin-apps-owner AT cygwin DOT com]On Behalf Of Pavel Tsekov > > > Sent: Friday, February 15, 2002 10:05 AM > > > > > > Robert Collins wrote: > > > > > > > Ok, > > > > finally got some breathing time. > > > > > > > > Setup with char * eliminated is ~350K. Ouch. > > > > > > > > This is why I've not committed my patch yet (I've been trying to > see > > > > *where* the extra 100K appeared from). > > > > > > > > > You have four 'inline' - I know they're small in size, but three > of > them > > > are the most commonly used methods (the default and the copy > constructor > > > and also the 'operator ='). Remove the 'inline' modifier and see > if > the > > > executable gets smaller. > > > > > > > Done and done: > > > > CVS + "For the curious" patch + Two subsequent patches from Michael > Chase == > > 355840 bytes. > > Above with all inlines "un-inlined" == 344576 bytes. > > > > So a bit over 11KB saved. In my judgement that's enough to warrant > removing the > > inlines; if string-handling speed is a significant factor for > setup.exe I'd say > > there's something wrong somewhere that no amount of inlines could > remedy. > > > > -- > > Gary R. Van Sickle > > Brewer. Patriot. > > > > > >