Re: patch for postgresql driver

On Wed, Apr 17, 2013 at 08:28:21AM +0100, Wols Lists wrote:
> On 17/04/13 02:10, Lionel Elie Mamane wrote:
> > On Tue, Apr 16, 2013 at 09:24:03PM +0100, Wols Lists wrote:
> > > On 16/04/13 20:28, Lionel Elie Mamane wrote:

> > > > RTL_CONSTASCII_STRINGPARAM is a hack/optimisation specific to
> > > > compile-time constants. With your change, the string is not anymore
> > > > a compile-time constant

> > > I'm puzzled that it's not a compile-time constant

> > Well, it is not even a run-time constant: at each iteration of the
> > loop, it is a different value.

> Yes, the value passed changes each time, but the array itself is a
> constant array of constant strings ...

If you want to exploit that fact, you could do something like that:

  static std::vector< std::pair< char*, sal_Int32 > > keyword_list =
    boost::assign::list_of(::std::pair(RTL_CONSTASCII_STRINGPARAM( "password" )))
                          (::std::pair(RTL_CONSTASCII_STRINGPARAM( "user" )))
                          (etc)
                          (etc)
                          (etc).to_adapter()

  ....

      if( args[i].Name.matchIgnoreAsciiCaseAsciiL( keyword_list[j].first, keyword_list[j].second ))

(or for more C++-y style, use an iterator instead of indexed access)

Yes, C++03 makes this *much* harder than it could be (C++11 has a
better solution: http://en.wikipedia.org/wiki/C++11#Initializer_lists )

IMHO it is *far* more trouble than it is worth.

-- 
Lionel