Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder3.hal-mli.net!nx02.iad01.newshosting.com!newshosting.com!news-out.readnews.com!transit3.readnews.com!postnews.google.com!y39g2000prd.googlegroups.com!not-for-mail
From: kedar mhaswade <kedar.mhaswade@gmail.com>
Newsgroups: comp.lang.java.programmer
Subject: Re: Synchronization of the constructor
Date: Sat, 13 Aug 2011 23:12:56 -0700 (PDT)
Organization: http://groups.google.com
Lines: 63
Message-ID: <e46c0cd1-9906-4195-9d30-070318fa5318@y39g2000prd.googlegroups.com>
References: <j25hro$cjf$1@news.onet.pl> <j25uri$vm$1@dont-email.me> <j264b6$3u5$1@dont-email.me>
NNTP-Posting-Host: 208.127.246.23
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
X-Trace: posting.google.com 1313302376 22300 127.0.0.1 (14 Aug 2011 06:12:56 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: Sun, 14 Aug 2011 06:12:56 +0000 (UTC)
Complaints-To: groups-abuse@google.com
Injection-Info: y39g2000prd.googlegroups.com; posting-host=208.127.246.23; posting-account=lZExVgoAAAC6LNhvTOESNqEOAjRzDXhY
User-Agent: G2/1.0
X-Google-Web-Client: true
X-Google-Header-Order: HNKUARELSC
X-HTTP-UserAgent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/535.1 (KHTML, like Gecko) Ubuntu/10.10 Chromium/15.0.836.0 Chrome/15.0.836.0 Safari/535.1,gzip(gfe)
Xref: x330-a1.tempe.blueboxinc.net comp.lang.java.programmer:7101

On Aug 13, 8:13=A0am, markspace <-@.> wrote:
> On 8/13/2011 6:36 AM, Eric Sosman wrote:
>
> > On 8/13/2011 5:58 AM, MaciekL wrote:
> >> Hi,
>
> >> I have a doubt because Java disables synchronization of the constructo=
r
> >> by default.
>
> > The constructor can be synchronized,
>
> It really can't. =A0Recall that the compiler will insert a call to a supe=
r
> constructor if the first statement doesn't have a such a call or a call
> to another class constructor. =A0Consider this:
>
> public class SomeClass {
> =A0 =A0public SomeClass() {
> =A0 =A0 =A0synchronized( SomeClass.class ) {
> =A0 =A0 =A0 =A0...
> =A0 =A0 =A0}
> =A0 =A0}
> ...
>
> What you get is this:
>
> public class SomeClass {
> =A0 =A0public SomeClass() {
> =A0 =A0 =A0super();
> =A0 =A0 =A0synchronized( SomeClass.class ) {
> =A0 =A0 =A0 =A0...
> =A0 =A0 =A0}
> =A0 =A0}
> ...
>
> And thus you see that some writes in the construction occur outside of
> the synchronized block, a classic case of incorrectly written
> synchronization.
>
> The closest Java gets to a synchronized constructor is immutable objects
> made with final fields.
>
> public class Immutable {
> =A0 =A0private final SomeObject o;
> =A0 =A0public Immutable() {
> =A0 =A0 =A0o =3D new SomeObject();
> =A0 =A0}
> ...
>
> This is thread safe, and immutable, because the fields written are
> declared "final." =A0Java takes special processing at the end of the
> constructor to synchronize all final fields, and any writes made to
> objects accessible via those final fields, with all other threads in the
> system. =A0So now this Immutable class can be used safely by any thread i=
n
> the system.

Doesn't that depend on how Immutable SomeObject is and how Immutable's
immutability relates to it?
Also, other standard caveats apply (e.g. Immutable should be declared
final).