Tutorial for debugging LO with Visual Studio

[Regina Henschel]

Hi Takeshi Abe,


Takeshi Abe schrieb:
> Hi Regina,
>
> On Thu, 30 Jul 2015 23:24:43 +0200, Regina Henschel <rb.henschel at t-online.de> wrote:
>> I want to look why the MathML import has become so bad in LO (tdf#75171).
> I am also curious about tdf#75171.
> It is SmTextNode::CreateTextFromNode() that contains the code putting strange
> "italic" on single-letter identifiers when importing a MathML.
> Applying the attached patch will stop the behavior (and thus make mmlimport
> test cases fail).

Unfortunately the problem is deeper. There should not be any quotes, 
because the element <mi> from MathML should be mapped to a variable in 
starmath, but actually it is mapped to a text. Removing the italic style 
cured only a symptom, the element has still the wrong font and is the 
wrong type.

I have compared AOO and LO. They are identic along this way: <mit> --> 
XML_MI --> XML_TOK_MIi --> CreateIdentifierContext --> 
SmXMLIdentifierContext_Impl.  There the member nGroup is explicitely set 
to 0 in AOO and in LO this step is missing.

I want to look, whether nGroup and the other members have correct 
values. Until now I have done such things using printf, but I thought it 
would be good to learn how to use Visual Studio for such purpose.

> But I have little idea about why LibO 4.1 was OK while 4.2 and later are NG.

There had been a lot of changes by Frédéric Wang in starmath in quick 
succession at that time. I have already looked at them, but not seen 
anything obvious.

Kind regards
Regina