Re: [PATCH] tools/perf,tools/lib/traceevent: Make traceevent APIs more consistent
From: Steven Rostedt
Date: Fri Mar 22 2019 - 10:09:11 EST
On Fri, 22 Mar 2019 15:08:16 +0200
Tzvetomir Stoyanov <tstoyanov@xxxxxxxxxx> wrote:
> Rename few traceevent APIs, in order to make it more consistent:
> tep_pid_is_registered(), tep_file_bigendian(),
> tep_is_latency_format(), tep_get_header_page_ts_size(),
> tep_set_host_bigendian(), tep_is_host_bigendian() and
> tep_data_lat_fmt()
Note, change logs should be about "why" not "what". The above is mostly
about "what" although you did state "to make it more consistent" which
is a "why", but we should elaborate more. Something like this:
=====================================
Rename some traceevent APIs for consistency:
tep_pid_is_registered() to tep_is_pid_registered()
as the convention is "tep_is" not "tep_X_is".
tep_file_bigendian() to tep_is_file_bigendian()
as boolean operations will now have "tep_is_" notation
tep_get_header_page_ts_size() to tep_get_header_timestamp_size()
as the latter is more descriptive.
tep_set_host_bigendian() to tep_set_local_bigendian()
tep_is_host_bigendian() to tep_is_local_bigendian()
because "host" can be confused with VMs, and "local" is about the local
machine.
tep_host_bigendian() to tep_is_bigendian()
Again "host" is confusing and we are converting to "tep_is_" notation,
also this one checks the actual machine that is running. For
consistency we have:
bool tep_is_bigendian(void);
bool tep_is_file_bigendian(struct tep_handle *tep);
bool tep_is_local_bigendian(struct tep_handle *tep);
Where tep_is_X_bigendian(tep) returns the saved data in the tep handle,
and tep_is_bigendian() returns the running machine's endianess.
tep_data_lat_fmt() to tep_data_latency_format()
No need to obfuscate with shorten names here.
Switching all "tep_is_" functions to return bool and not int.
=====================================
That is much more descriptive of the change, and lets anyone that looks
at the git history see exactly why things have changed. With a more
descriptive change log, it will cause less arguments as well.
Please resend with an updated change log.
-- Steve