Re: [PATCH 0/2] libtraceeval: Remove the need of TRACEEVAL_TYPE_NONE

[Ross Zwisler <zwisler@google.com>]

On Mon, Oct 02, 2023 at 03:07:49PM -0600, Ross Zwisler wrote:
> On Fri, Sep 29, 2023 at 06:00:23AM -0400, Steven Rostedt wrote:
> > From: "Steven Rostedt (Google)" <rostedt@goodmis.org>
> > 
> > The use of macros to determine the size of the keys and vals arrays made
> > me think a bit more about how we initialize the traceeval. The more I'm
> > using the library, the less I like having to add the TRACEEVAL_TYPE_NONE
> > to the end of the keys and vals array.
> > 
> > Instead, use the same macro trick of TRACEEVAL_ARRAY_SIZE() to determine
> > the number of elements. Still allow the use of using the
> > TRACEEVAL_TYPE_NONE to determine the size (if the number passed in is
> > still bigger), so that most applications that still use that still work.
> > 
> > I made the change separate than updating the sample code to test that
> > the old way still works. Then I updated the sample to make sure the new
> > way works. Perhaps when we add unit tests back in, we'll test both
> > cases.
> 
> Looks good, you can add:
> Reviewed-by: Ross Zwisler <zwisler@google.com>

Actually, I think we need to make similar updates in check_keys() and
check_vals(), else calls to traceeval_init_data_size() will walk off the end
of those respective arrays.

> 
> > Steven Rostedt (Google) (2):
> >   libtraceeval: Remove need to use TRACEEVAL_TYPE_NONE in keys and vals
> >   libtraceeval samples: Remove adding TRACEEVAL_TYPE_NONE to keys and
> >     vals
> > 
> >  include/traceeval-hist.h | 11 +++++++++--
> >  samples/task-eval.c      | 18 ------------------
> >  src/histograms.c         | 11 +++++++----
> >  3 files changed, 16 insertions(+), 24 deletions(-)
> > 
> > -- 
> > 2.40.1
> >