Date: Thu, 11 Nov 2004 14:55:12 +0100
From: Jan Hudec <bulb@ucw.cz>
To: Catalin Marinas <catalin.marinas@arm.com>
Message-ID: <20041111135511.GZ721@vagabond>
References: <20041107234609.7bf0abfe@delta.hk.office.outblaze.com>
	<tnxlldcbqm1.fsf@arm.com>
	<D4D5B1A3-31DC-11D9-B0B3-000393CFE6B8@spy.net>
	<1099995711.2900.84.camel@stargate>
	<20041110204049.GD5978@suffields.me.uk>
	<20041111093245.GY721@vagabond> <tnxbre4pqgl.fsf@arm.com>
Mime-Version: 1.0
In-Reply-To: <tnxbre4pqgl.fsf@arm.com>
User-Agent: Mutt/1.5.6+20040907i
Cc: gnu-arch-users@gnu.org
Subject: [Gnu-arch-users] Re: darcs vs tla
Precedence: list
Content-Type: multipart/mixed; boundary="===============1099442541=="
Sender: gnu-arch-users-bounces+migo=homemail.com@gnu.org
Errors-To: gnu-arch-users-bounces+migo=homemail.com@gnu.org
Status: RO
Content-Length: 3001
Lines: 85

--===============1099442541==
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature"; boundary="Qesg98+uAFykT7J8"
Content-Disposition: inline


--Qesg98+uAFykT7J8
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Nov 11, 2004 at 11:36:42 +0000, Catalin Marinas wrote:
> Jan Hudec <bulb@ucw.cz> writes:
> > There is an interesting paper by Paul Graham. He has a hypothesis, that
> > a programmer can write about the same number of tokens in a unit of
> > time, no matter what language he is writing them in (or little
> > matter).
>=20
> That's only useful if the first thing you do when starting a project
> is coding. It might be irrelevant for large projects where you begin
> with a design stage. I think it's wrong to estimate the cost of a
> projects by counting the number of lines.

For a big part, the design cost is independent on the language. The
higher layers of design are same independent of the language.

But the lower level split to modules is proportional to size of
interfaces. The more expressive interfaces the language allows, the less
work is designing them. Expressivnes of interfaces is usualy tied to
expressivness of the language as a whloe, so here again HLLs have
advantage, though not that big.

> > Now that would mean, that programming is more efficient in a higher
> > language. The higher here means it can do more work with one
> > statement.
>=20
> C is a language where you might have to implement a lot of things
> yourself. In C++, things are a bit different. You need a list, just
> use the STL or some other template library (Boost). If, after
> profiling your code, you realise that this list is slow you can easily
> change it (providing that you had some rules like using typedefs
> etc.). You might not have this option in a HLL.

Replacing eg. STL list with a Boost one rarely wins much. What wins much
is replacing eg. list with a vector -- generaly with a different
structure with more appropriate asymptotic behaviour. Reasonable HLLs
give you this choice.

---------------------------------------------------------------------------=
----
						 Jan 'Bulb' Hudec <bulb@ucw.cz>

--Qesg98+uAFykT7J8
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iD8DBQFBk28/Rel1vVwhjGURAhWPAJ4qHbX1tKH6q8lR45C2W/R9cHak3QCbBeuH
fFyO4xcfj9v9cgx4GczbWKQ=
=OESr
-----END PGP SIGNATURE-----

--Qesg98+uAFykT7J8--


--===============1099442541==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Gnu-arch-users mailing list
Gnu-arch-users@gnu.org
http://lists.gnu.org/mailman/listinfo/gnu-arch-users

GNU arch home page:
http://savannah.gnu.org/projects/gnu-arch/
--===============1099442541==--