[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20170731153539.255ab561@gandalf.local.home>
Date: Mon, 31 Jul 2017 15:35:39 -0400
From: Steven Rostedt <rostedt@...dmis.org>
To: Federico Vaga <federico.vaga@...a.pv.it>
Cc: LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 2/2] trace-cmd: replace show_file() ->
show_instance_file()
On Mon, 10 Jul 2017 02:15:35 +0200
Federico Vaga <federico.vaga@...a.pv.it> wrote:
> On Friday, July 7, 2017 12:29:35 AM CEST Steven Rostedt wrote:
> > On Mon, 5 Jun 2017 11:31:18 +0200
> >
> > Federico Vaga <federico.vaga@...a.pv.it> wrote:
> > > show_file(name) and show_instance_file(&top_instance, name) are
> > > equivalent.
> > >
> > > Remove the show_file() function in order to have a single function for
> > > this task.
> >
> > Actually I find nothing wrong with having a helper function like this.
> > IIRC, show_file() was first, and then show_instance_file() came later.
> > There's some files that only exist for the top_instance, and I like the
> > fact that this is annotated that way.
> >
> > I'm curious to know what the benefit of removing show_file() is?
>
> The show_file(name) and show_instance_file(&top_instance, name) are
> equivalent: they do the same thing. By removing `show_file` the developers are
> forced to use always the same function and being explicit about the instance
> they want to use.
>
> The name `show_file()` is so generic that does not implies automatically that
> we are accessing the top_instance. This is not even clear by reading the
> implementation; people must read the other functions used in `show_file()` to
> understand that their instance scope is always 'top_instance'.
>
> So, in my opinion, it makes the code easier to read and more explicit in what
> is doing without too much effort.
>
Just an FYI. You'll find lots of these types of helper functions in the
Linux Kernel. As I'm a Linux Kernel developer, I prefer them ;-)
-- Steve
Powered by blists - more mailing lists