RFR: (8030875) Macros for checking and returning on exceptions

[Martin Buchholz]

On Wed, Jan 15, 2014 at 12:52 PM, roger riggs <roger.riggs at oracle.com>wrote:

>  Hi Martin,
>
> A followup, what is the subtlety that makes it beneficial to wrap
> even a single statement in a do {} while (0)?
>
>
It ensures that a macro call FOO() has the same semantics as a regular
non-macro void function call, e.g.
FOO()
is an unfinished statement until the terminating semicolon is added.


> It makes sense for macros with multiple statements but not for a single
> statement substitution of expressions.
>
> For a single statement (w/o ';'), it seems like a redundant overhead
> that does not solve an issue.
>
>
If your compiler cannot optimize away
 while (false)
it might be time to get a new one.


> Thanks, Roger
>
>
> On 1/14/2014 5:20 PM, Martin Buchholz wrote:
>
>
>
>
> On Tue, Jan 14, 2014 at 7:54 AM, roger riggs <roger.riggs at oracle.com>wrote:
>
>> Hi David,
>>
>> The CHECK_RETURN macros have existed in java.net for some time and
>> I have not seen any empty statement warnings.
>>
>> The CHECK_EXCEPTION macros are new and does not have any uses yet.
>> I don't plan to do any wholesale modification of current sources.
>>
>> The macros always produce a valid statement;  (though my c/c++ may be a
>> bit rusty).
>>
>>
>  Macros should not generally be complete statements, including
> semicolons; then the source code looks like this:
>
>  FOO()
>
>  which looks (including to emacs) like someone forgot the trailing
> semicolon.
>
>  Instead, these macros should be written using the "well-known" do ( ...
> ) while (0) idiom, e.g.
>
>  #define CHECK_NULL(x) do { if ((x) == NULL) return; } while (0)
>
>
>> (Note that it has been requested to rename the macros to include a
>> JNU_prefix
>> to be consistent with other macros in jni_util.h. I will propose a
>> separate set of changes for that).
>>
>> Roger
>>
>>
>> On 1/13/2014 9:51 PM, David Holmes wrote:
>>
>>> Hi Roger,
>>>
>>>
>>> webrev:
>>>> http://cr.openjdk.java.net/~rriggs/webrev-check-exception-8030875/
>>>>
>>>
>>> Do these macro definitions not cause "empty statement" warnings from the
>>> compiler? (Existing ones have the same problem I guess)
>>>
>>> Also I don't see any use of the CHECK_EXCEPTION macros? In fact the bulk
>>> of changes here seem completely unrelated to the "exception" aspect of this
>>> CR.
>>>
>>> Cheers,
>>> David
>>>
>>> [1] https://bugs.openjdk.java.net/browse/JDK-8030875
>>>>
>>>
>>
>
>