Hi Andy,<br><br><div class="gmail_quote">On Tue, Jun 12, 2012 at 7:01 AM, Andy Gibbs <span dir="ltr"><<a href="mailto:andyg1001@hotmail.co.uk" target="_blank">andyg1001@hotmail.co.uk</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Attached is a patch fixing a preprocessor paste bug.<br></blockquote><div><br></div><div>Great, thanks!</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
The bug manifests itself by swallowing up the remainder of the file<br>
following an invalid paste operation. In terms of technical details,<br>
TokenLexer::PasteTokens was returning tok::eof in this case, effectively<br>
stopping any further processing. The bug includes a test-case demonstrating<br>
the fix. No other regressions are noted.</blockquote><div><br></div><div>In -fms-extension -Wno-invalid-token-paste mode, the patch regresses this test, because we're no longer invoking LookUpIdentifierInfo on the pasted 'ab' token:</div>
<div><br></div><div> #define foo a ## b ## = 0</div><div> int foo;</div><div><br></div><div>Switching out the 'return true' for a 'break' should resolve that, I think. Your fix should also apply in the AsmPreprocessor case.</div>
<div><br></div><div>Can you merge your test case into the existing file?</div></div>