[cfe-commits] r159939 - in /cfe/trunk: docs/ReleaseNotes.html include/clang/Basic/DiagnosticSemaKinds.td lib/Sema/SemaExpr.cpp test/FixIt/objc-literals.m test/SemaObjC/objc-literal-comparison.m

[Jean-Daniel Dupas]

Or maybe as a side note somewhere in the Obj-C literals doc at http://clang.llvm.org/docs/ObjectiveCLiterals.html

This document is supposed to "describes how the features are implemented in clang, and how to use them in your own programs." so it is probably a good place to discuss such implementation detail.

Le 9 juil. 2012 à 19:09, Jordan Rose a écrit :

> Hm. It's a bit much to put in the actual warning text, though, and most people who use clang for Objective-C won't be looking at Clang release notes anyway. Maybe I can try to put it through in our internal documentation (what appears on developer.apple.com).
> 
> Jordan
> 
> 
> On Jul 9, 2012, at 10:06 , Jean-Daniel Dupas <devlists at shadowlab.org> wrote:
> 
>> While this is just a warning now, I found the comment in the release note very helpful to understand why we should never do it.
>> 
>> Le 9 juil. 2012 à 18:54, Jordan Rose a écrit :
>> 
>>> Author: jrose
>>> Date: Mon Jul  9 11:54:44 2012
>>> New Revision: 159939
>>> 
>>> URL: http://llvm.org/viewvc/llvm-project?rev=159939&view=rev
>>> Log:
>>> Downgrade the "direct comparison" error for ObjC literals to a warning.
>>> 
>>> Chris pointed out that while the comparison is certainly problematic
>>> and does not have well-defined behavior, it isn't any worse than some
>>> of the other abuses that we merely warn about and doesn't need to make
>>> the compilation fail.
>>> 
>>> Revert the release notes change (r159766) now that this is just a new warning.
>>> 
>>> Modified:
>>>  cfe/trunk/docs/ReleaseNotes.html
>>>  cfe/trunk/include/clang/Basic/DiagnosticSemaKinds.td
>>>  cfe/trunk/lib/Sema/SemaExpr.cpp
>>>  cfe/trunk/test/FixIt/objc-literals.m
>>>  cfe/trunk/test/SemaObjC/objc-literal-comparison.m
>>> 
>>> Modified: cfe/trunk/docs/ReleaseNotes.html
>>> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/docs/ReleaseNotes.html?rev=159939&r1=159938&r2=159939&view=diff
>>> ==============================================================================
>>> --- cfe/trunk/docs/ReleaseNotes.html (original)
>>> +++ cfe/trunk/docs/ReleaseNotes.html Mon Jul  9 11:54:44 2012
>>> @@ -218,21 +218,7 @@
>>> <h3 id="objcchanges">Objective-C Language Changes in Clang</h3>
>>> <!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
>>> 
>>> -<ul>
>>> -  <li>
>>> -    <p>It is now an error to compare against the addresses of Objective-C
>>> -       literals. This is usually a simple mistake (using <code>==</code> instead
>>> -       of <code>-isEqual:</code>), and the result depends on the implementation
>>> -       of the various literals, none of which are guaranteed to be uniqued or
>>> -       always newly-allocated.</p>
>>> -    <p>In the past, we allowed comparisons against literal strings
>>> -       (<code>@"..."</code>), since they are currently uniqued across
>>> -       translation units at link time. This is an implementation detail and
>>> -       should not be relied upon. If you are using such code, please use global
>>> -       string constants instead (<code>NSString * const MyConst = @"..."</code>)
>>> -       or use <code>-isEqual:</code>.</p>
>>> -  </li>
>>> -</ul>
>>> +<p>...</p>
> 

-- Jean-Daniel