[cfe-dev] Usefull warnings

Fernando Pelliccioni fpelliccioni at gmail.com
Mon Apr 21 08:07:34 PDT 2014


Sorry!

lock = look

Please, don't look the attached file using Google Preview. Download it and
open it with your favorite text editor.


On Mon, Apr 21, 2014 at 12:06 PM, Fernando Pelliccioni <
fpelliccioni at gmail.com> wrote:

> Please, don't lock the attached file using Google Preview. Download it and
> open it with your favorite text editor.
>
>
> On Mon, Apr 21, 2014 at 12:04 PM, Fernando Pelliccioni <
> fpelliccioni at gmail.com> wrote:
>
>> Hi all,
>>
>> The following code snippet is related to the Apple bug of a few weeks ago.
>>
>> void foo();
>> void bar();
>>
>> void test( int a )
>> {
>> if ( a == 1 )
>> foo();
>> bar();
>> }
>>
>> The original source code of the Apple bug has Goto's instead of function
>> calls, but the problem is essentially the same, and it is related to
>> scoping.
>>
>> I wonder if you considered convenient adding a warning if Clang detects
>> that a line is indented so that does not match the natural indentation of
>> its scope, especially there are a conditional around.
>>
>>
>> The following code is fine, no problem with it.
>> void test2()
>> {
>> foo();
>> bar();
>> }
>>
>>
>> I think the followings cases should be warned by the compiler.
>> void test3()
>> {
>> while (true) foo();
>>  bar();
>> }
>>
>> void test4()
>> {
>> if (true) foo();
>> bar();
>> }
>>
>> To emphasize the problem I attached a source code file. Could you give it
>> a quick look and tell me what is printed out?
>> I know that it is an extreme and malicious example, but it help us to
>> understand that when we given a quick look to the code, we can ignore cases
>> like these.
>>
>>
>> Do you think it would be useful to add these warnings to Clang?
>>
>> Thanks and regards,
>> Fernando Pelliccioni.
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20140421/6c566b94/attachment.html>


More information about the cfe-dev mailing list