Hi Niklas
While not involved with reference parsing, I do know that for say the
example range like "Sheet1!A1" currently code that Jody popped into the
vba interop api to help me ( for various range XL/R1C1 support )
expects SCA_VALID to be set on return
>From your comments below it would seem you imply SC_VALID getting set
for a single reference is not VALID ( pardon the pun ), and that causes
subsequent problems correct?
The parsing code ( or at least the call to the parse code ) that Jody
added for me always seems to expect SCA_VALID to be set ( and indeed
it's also passed as the mask )
e.g.
USHORT nMask = SCA_VALID;
rResFlags = rCellRanges.Parse( sAddress, pDoc, nMask, eConv, 0
);
if ( rResFlags & SCA_VALID )
return true;
return false;
So with respect to the SCA_VALID flag, I can live with a different
return value that can indicate success. But I would worry though that
SC_VALID is indeed required in a non obvious way ( there are many
strange xl addressing edge cases ), care would be required I think
changing this.
I could test any changes against my current ( not exhaustive but better
than nothing ) addressing tests ( that at least exercise this code a
little ) p.s. this week is not good though
For the other scenario you mention nothing immediately comes to mind.
Noel
On Mon, 2007-06-25 at 18:45 +0200, Niklas Nebel wrote:
> Hi everyone who's involved with reference parsing,
>
> When parsing references with CONV_XL_A1 convention, the return values
> from ScRange::Parse are different from those for CONV_OOO and don't
> match what ScCompiler::IsReference expects.
>
> - For a single reference (Sheet1!A1), the bit SCA_VALID is set, which
> causes a ScDoubleRefToken to be generated.
>
> - For a range reference like "Sheet1!A1:A2" (with only one sheet name),
> SCA_VALID_TAB2 isn't set, which leads to an invalid reference.
>
> Is there any specific reason for this difference? Is it needed somewhere
> (some code not using ScCompiler, or some private changes to it)?
> Otherwise, I'm going to change it, returning no SCA_VALID bit in the
> first case, and SCA_VALID_TAB2 (but not SCA_TAB2_3D) in the second case.
>
> Niklas
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
--
Noel Power
OpenOffice Developer
Novell Ireland Software Limited
Corrig Court, Corrig Road
Sandyford Business Estate, Sandyford
Dublin 18
[EMAIL PROTECTED]
Novell, Inc.
Software for the Open Enterprise
www.novell.com/open
SUSE® Linux Enterprise 10
Your Linux is ready
www.novell.com/linux
Registered No: 223694
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
