Re: [PATCH 01/15 v3] Introduce common/errors.h

[Pedro Alves <palves@redhat.com>]

On 07/17/2014 04:39 PM, Gary Benson wrote:
> Pedro Alves wrote:
>> On 07/17/2014 02:47 PM, Gary Benson wrote:
>>> +/* Throw an error.  The current operation will be aborted.  The
>>> +   message will be issued to the user.  The application will
>>> +   return to a state where it is accepting commands from the user.  */
>>
>> These comments aren't really true, though.  We have plenty of places
>> catching errors with TRY_CATCH and proceeding without returning to a
>> state where we're accepting commands from the user.
> 
> How about "Throw an error.  The current operation will be aborted.
> The application may catch and process the error, or, if not, the
> message will be issued to the user and the application will return
> to a state where it is accepting commands from the user."

Still infers what the application does with the error.  IMO, it's
best to describe the interface.  Like:

 Throw a generic error.  This function not return, instead
 it executes a long jump, aborting the current operation.
 The function takes printf-style arguments, which are used to
 construct the error message.

>> (We should really bite the bullet and move exceptions.{h|c} and
>> cleanups.{h|c} to common/ and make gdbserver use them too.)
> 
> Can I leave this for another series please?  This one's big enough
> already.

Yes, of course.  That's why it was a parenthesis.

-- 
Pedro Alves