Re: [PATCH] Move "tee" building down to interpreter::set_logging_proc

[Simon Marchi <simon.marchi@polymtl.ca>]

On 2017-02-02 12:39, Pedro Alves wrote:
> On 02/02/2017 03:17 PM, Simon Marchi wrote:
>> On 2017-02-02 09:28, Pedro Alves wrote:
>>> @@ -109,13 +109,13 @@ extern int current_interp_named_p (const char
>>> *name);
>>> 
>>>  /* Call this function to give the current interpreter an opportunity
>>>     to do any special handling of streams when logging is enabled or
>>> -   disabled.  START_LOG is 1 when logging is starting, 0 when it 
>>> ends,
>>> -   and OUT is the stream for the log file; it will be NULL when
>>> -   logging is ending.  LOGFILE is non-NULL if the output streams
>>> +   disabled.  START_LOG is true when logging is starting, false when
>> 
>> START_LOG is not there anymore.  From what I understand, it's replaced
>> with LOGFILE being null or not?
> 
> You're right.  How about this:
> 
> -- i/gdb/interps.h
> +++ w/gdb/interps.h
> @@ -109,10 +109,10 @@ extern int current_interp_named_p (const char 
> *name);
> 
>  /* Call this function to give the current interpreter an opportunity
>     to do any special handling of streams when logging is enabled or
> -   disabled.  START_LOG is true when logging is starting, false when
> -   it ends.  LOGFILE is the stream for the log file; it's NULL when
> -   logging is ending.  LOGGING_REDIRECT is false if the output streams
> -   are to be tees, with the log file as one of the outputs.  */
> +   disabled.  LOGFILE is the stream for the log file when logging is
> +   starting and is NULL when logging is ending.  LOGGING_REDIRECT is
> +   false if the output streams are to be tees, with the log file as
> +   one of the outputs.  */
> 
>  extern void current_interp_set_logging (ui_file_up logfile,
>                                         bool logging_redirect);
> 
> 
> OK?

Yeah sounds good.

Though the pre-existing sentence "...if the output streams are to be 
tees" is not that clear to me, I'm not sure I would understand if I 
didn't already know what the function does.  Why does it talk about 
multiple output streams that have to be tees, isn't there only one tee?  
Or is it meant to be a past tense verb, in which case it should be 
something like "...are to be tee-ed"?  I just find the formulation 
awkward.