From: <telford AT xenon DOT triode DOT net DOT au>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: [Q] Size of compiled executables from GCC 2.7x vs. 2.8
Date: 21 Apr 1998 02:05:55 GMT
Organization: Triode Internet
Lines: 31
Message-ID: <6hguu3$923$3@hyperion.triode.net.au>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 980419142226 DOT 23362P-100000 AT is>
NNTP-Posting-Host: xenon.triode.net.au
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 8bit
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

In comp.os.msdos.djgpp Eli Zaretskii <eliz AT is DOT elta DOT co DOT il> wrote:

> On Fri, 17 Apr 1998, Anshuman Pandey wrote:

> > I noticed that my executables were considerably larger when compiled with
> > gcc v2.8 and bnu v2.81 than they were originally when compiled with gcc 
> > v2.721 and v2.7. The increase in size was about 100k. Does anyone happen
> > to know what the reason behind this might be?

> You didn't tell enough to answer that.  First, are those programs
> stripped?  In other words, did you run `strip' on them, or used -s
> option when linking?  If not, they include debugging info which might
> be larger in later versions (more debugging info generally means
> better debugging).

> If they are stripped, you need to tell which switches did you use when
> compiling.  Some switches can have different effects with different
> compiler versions.  For example, if you use -O3, the amount of code
> inlined by the compiler could change with the versions.

> Also, how large are the programs?  Some bloat might be due to
> different alignment of the code/data.

> In short, tell more.

I'm guessing that they have exception tables in them which take
up bulk space and don't get removed by strip and are only useful
if you actually use exceptions in your C++ programs.

	- Tel