proper format for printing GetLastError()

On Mon, Jul 4, 2011 at 17:29, Peter Eisentraut <peter_e@gmx.net> wrote:

About half of our code prints GetLastError() using %d after casting it
to int (actually, about half of that half uses %i, another thing to sort
out, perhaps), and the other half uses %lu without casting.  I gather
from online documentation that GetLastError() returns DWORD, which
appears to be unsigned 32 bits.  So using %lu appears to be more
correct.  Any arguments against standardizing on %lu?

Nope. I think it's only in there because of lazyness, in general. %lu
seems to be the correct choice.

Secondly, it might also be good if we could standardize on printing

   actual message: error code %lu

instead of just

   actual message: %lu

Or "actual error code: %lu"?

Thirdly, why are we not trying to print a textual message?

I'd say that depends on where it is. In some cases probably because
it's "can never happen" messages. In other cases because, well, no
reason :)

--
 Magnus Hagander
 Me: http://www.hagander.net/
 Work: http://www.redpill-linpro.com/

Yeah, I noticed that myself recently.

On 6 July 2011 12:48, Magnus Hagander <magnus@hagander.net> wrote:

Nope. I think it's only in there because of lazyness, in general. %lu
seems to be the correct choice.

Yes, it's the correct choice.

Thirdly, why are we not trying to print a textual message?

I'd say that depends on where it is. In some cases probably  because
it's "can never happen" messages. In other cases because, well, no
reason :)

I'd also say it has something to do with the win32 API being as ugly
and counter-intuitive as it is. Take a look at this:

http://msdn.microsoft.com/en-us/library/ms680582(v=VS.85).aspx

--
Peter Geoghegan       http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training and Services