Re: [smartmontools-support] Next Windows Release?

[Andreas M. <an...@us...>]

Hi,

On Sun, Jan 13, 2013 at 02:09:29AM +0000, sma...@li... wrote:
> Date: Sat, 12 Jan 2013 16:29:34 -0600
> From: Kyl Scott <kyl...@ti...>

Let me just say that this email was perfect (I especially liked the
"idiot" part, but I'm sure the controller details might be helpful
to some people as well).

> Whereupon I read the MSDN description which pointed out that little 
> detail about having to be running in adminstrator mode for it to work 
> with win7/vista and GENERIC_WRITE...
> 
> So starting my command window in administrator mode allows the latest 
> official release to work just fine and the 1882ix is now recognized.  
> Umm, yeah, I did mention idiot... ;)
> 
> So the question I am wondering about is whether it makes sense to add a 
> message (or maybe even a permission check) to smartctl about 
> administrator mode on windows and warn newcomers like me when their 
> requested controller is not found due to improper invocation?  This 
> requirement is probably in the docs somewhere, but I missed it when I 
> was skimming over things (still not used to UAC, at all).  Not a big 
> deal either way, but I thought I would toss the idea out there.  If it 
> is a foolish one, then please freely ignore it. ;)

It seems like such a warning would be very desireable from a usability POV.

Rather than doing a broad admin-permissions check and always printing
such a warning on permission failure (which might or might not be
correct about a request actually subsequently failing due to permissions),
which would be a rather unspecific way of checking this,
it should probably be implemented in the most direct way possible.
I.e. directly check for request failure in Win32 code layer,
and generate such a warning message, either by generating it right in
the request handling layer (dirty), or perhaps signalling permission
error to upper smartmontools handling via a smartmontools-specific error enum,
for upper handling to then be able to generate an appropriate error.

Possible pseudo code:
  res = do_request_inner_win32();
  if (failed(res) && perms_failure_probable(res))
  {
    if (insufficient_perms_win32() /* do UAC checks and such */)
      res = SMARTMON_INSUFFICIENT_PERMS;
  }

If nobody is (avail~)able to fully implement such a check currently,
then at least a prominent FIXME comment should be added
immediately.

Andreas Mohr