Path: csiph.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail
From: Robert Klemme <shortcutter@googlemail.com>
Newsgroups: comp.lang.java.programmer
Subject: Re: Type conversions with the empty string
Date: Mon, 30 Jan 2017 10:08:30 +0100
Lines: 76
Message-ID: <ef8e4gFgaq3U1@mid.individual.net>
References: <emtpy-string-20170128143107@ram.dialup.fu-berlin.de> <ef6l8kF5p6bU1@mid.individual.net> <gkienx5o72gf.r7zyv6z3bfpc.dlg@40tude.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: quoted-printable
X-Trace: individual.net T+TE2XiqjmmqFM5O1S/KCgOrbr8FD0bqTXUN0pvRM9oH6RXb4=
Cancel-Lock: sha1:i7Kds8G5PJSbY0tWfFslBaarRdk=
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.7.0
In-Reply-To: <gkienx5o72gf.r7zyv6z3bfpc.dlg@40tude.net>
Xref: csiph.com comp.lang.java.programmer:36224

On 30.01.2017 07:39, Joerg Meier wrote:
> On Sun, 29 Jan 2017 17:57:55 +0100, Robert Klemme wrote:
>
>> On 28.01.2017 14:43, Stefan Ram wrote:
>>>   I'd say that
>
>>> "".valueOf( 123 )
>
>>>   is even more robust than
>
>>> String.valueOf( 123 )
>
>>>   because the former is still working when a file =C2=BBString.class=C2=
=AB
>>>   is found early enough in the class path.
>> What do you mean by that?  Why would the second form be less robust?
>> This does not make any sense to me as the same class is used.
>
> Check http://ideone.com/9xXYpi
>
> 		System.out.println("" + 123);
> 		System.out.println(String.valueOf(123));
> 	=09
> ends up printing out:
>
> 123
> 15129

Your remark about class path mislead me.  This is not about class path=20
but about namespaces.

Come on.  This is a constructed edge case for various reasons:

  - Although it is of course perfectly legal to implement a class called =

"String" in any other package someone who does that would have to have=20
very good reasons to do it - especially since it makes any reference to=20
java.lang.String extremely tedious.
  - Even more so as it shadows methods in java.lang.String with methods=20
having the same name but incompatible semantics.
  - PrintStream.println(int) does exist, so would have been the much=20
more reasonable and concise choice which does not fall into this trap.

https://docs.oracle.com/javase/8/docs/api/java/io/PrintStream.html#print-=
int-

Basically you are using a String constant as a shortcut for the full=20
qualified class name java.lang.String.

Btw. that class could still have used java.lang.String.valueOf() and=20
avoided the String constant:

class String {
   public static java.lang.String valueOf(int i) {
     return java.lang.String.valueOf(i * i);
   }
}

For me this edge case is so obscure that it does not warrant digressing=20
from what a reasonable programmer would do and to avoid the use of a=20
String constant here.

This case does exhibit a interesting detail of the Java compiler in=20
Eclipse Mars 4.5.2: even though invocation of static methods on=20
instances is set to be flagged as warning it does /not/ bark on the line =

with "" + 123.  That is odd.

Kind regards

	robert

--=20
remember.guy do |as, often| as.you_can - without end
http://blog.rubybestpractices.com/