From mboxrd@z Thu Jan  1 00:00:00 1970
From: Marek Vasut <marex@denx.de>
Date: Sun, 15 Sep 2013 16:21:08 +0200
Subject: [U-Boot] [PATCH v3] usb: new board-specific USB init interface
In-Reply-To: <5229C7E8.9070803@samsung.com>
References: <1375786242-11734-1-git-send-email-m.zalega@samsung.com>
	<201309061340.32372.marex@denx.de> <5229C7E8.9070803@samsung.com>
Message-ID: <201309151621.08392.marex@denx.de>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

Dear Mateusz Zalega,

> On 09/06/13 13:40, Marek Vasut wrote:
> >>>>>>> Moreover, the 'int index' should likely be unsigned int and the
> >>>>>>> special value to init all controllers at once should probably then
> >>>>>>> be 0xffffffff
> >>>>>> 
> >>>>>> Despite our greatest ambitions, I don't think we're likely to use
> >>>>>> more than 2^31-1 USB controllers at a time. Besides, negative
> >>>>>> values look better both in code and debugger session.
> >>>>> 
> >>>>> Thinking of it further, instead of using negative value here, like I
> >>>>> mentioned above, why not make the "board_usb_init_type" into a field
> >>>>> of flags , then add flag to init all controllers at once ?
> >>>> 
> >>>> That's unnecessary. It wouldn't lead to any practical advantage over
> >>>> existing interface.
> >>> 
> >>> The advantage would be you won't be mixing two things (value AND value
> >>> with special meaning) into the "index" parameter.
> >> 
> >> Alright, provide a use-case. The only 'special' value we have now
> >> doesn't interfere with controller index. Why write code or interfaces
> >> that won't ever be used?
> > 
> > Look, abusing the index field with a special value is moronic, especially
> > if you
> 
> I wouldn't call a de-facto standard abusive or moronic.
> On the other hand, contributing to unnecessary code bloat would be.
> 
> > _do_ have another field that can very well be turned into a block of
> > flags
> 
> Have you provided another use-case for that, as I asked?
> 
> > instead of enum right next to it.
> > 
> > function(int foo , enum bar)
> > 
> >             |         ^
> >             
> >             `---------'
> >             
> >              flags go
> >              
> >                here
> > 
> > Then int foo can be turned into unsigned int foo _and_ be used for it's
> > one singular purpose. Likewise, enum bar will now be "unsigned int
> > flags" . Do you see the separation now ?
> 
> Read your mail carefully - at this point I don't have any problems
> understanding what you wish to do, but I see your API changes as
> unjustified. Can you justify them without appealing to your authority
> and maybe inserting a couple of ad-hominems here and there?

I suppose this thread can be concluded by droping the INIT_ALL stuff entirely. 
Afterall, we do not want to init _ALL_ ports at once, but we want to init them 
selectively.

Best regards,
Marek Vasut