Provide better error message when passing wrong number of arguments.

Provide better error message when passing wrong number of arguments.
Committed: https://code.google.com/p/dart/source/detail?r=9602

[regis, 2012-07-11 00:35:28]

Srdjan,

Please, hold on on this review.
I need to fix some API tests doing string matching on the error message.
I think I also need to subtract the hidden arguments from the reported counts,
or the "improved" error messages could be actually more confusing than before.

Thanks,
Regis

[regis, 2012-07-11 23:22:22]

On 2012/07/11 00:35:28, regis wrote:
> Srdjan,
> 
> Please, hold on on this review.
> I need to fix some API tests doing string matching on the error message.
> I think I also need to subtract the hidden arguments from the reported counts,
> or the "improved" error messages could be actually more confusing than before.
> 
> Thanks,
> Regis

Srdjan, you can now have a look.

Thanks,
Regis

[srdjan, 2012-07-12 00:06:37]

LGTM with comments

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.c...
runtime/vm/object.cc:3715: if (message_buffer_size > 0) {
|| message_buffer == NULL

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.c...
runtime/vm/object.cc:3729: if (message_buffer_size > 0) {
ditto

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.c...
runtime/vm/object.cc:3774: if (message_buffer_size > 0) {
|| (message_buffer == NULL)

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.h
File runtime/vm/object.h (right):

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.h...
runtime/vm/object.h:1465: bool AreValidArgumentCounts(int num_arguments,
Add comment what message_buffer contains and that it can be NULL.

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.h...
runtime/vm/object.h:1473: 
Instead of passing the and allocating the message_buffer, it seems simpler to
return a String instead.

[regis, 2012-07-12 17:37:36]

Thanks!

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.c...
runtime/vm/object.cc:3715: if (message_buffer_size > 0) {
On 2012/07/12 00:06:37, srdjan wrote:
> || message_buffer == NULL

OS::SNPrint will work with a NULL buffer, as long as size is 0. So, an
ASSERT(message_buffer != NULL) if the size > 0 would prevent bad callers.

Since I will now use a String, this is not relevant anymore.

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.c...
runtime/vm/object.cc:3729: if (message_buffer_size > 0) {
On 2012/07/12 00:06:37, srdjan wrote:
> ditto

ditto

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.c...
runtime/vm/object.cc:3774: if (message_buffer_size > 0) {
On 2012/07/12 00:06:37, srdjan wrote:
> || (message_buffer == NULL)

ditto

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.h
File runtime/vm/object.h (right):

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.h...
runtime/vm/object.h:1465: bool AreValidArgumentCounts(int num_arguments,
On 2012/07/12 00:06:37, srdjan wrote:
> Add comment what message_buffer contains and that it can be NULL.

Done

https://chromiumcodereview.appspot.com/10695137/diff/2002/runtime/vm/object.h...
runtime/vm/object.h:1473: 
On 2012/07/12 00:06:37, srdjan wrote:
> Instead of passing the and allocating the message_buffer, it seems simpler to
> return a String instead.

Both functions now take a pointer to a String handle.