llvm dev - Aug 2009 - [LLVMdev] Buggy assertion in APFloat::convertFromHexadecimalString

Hello.

When running clang in Debug mode, I am getting an assertion failure when 
parsing the following line:

float ko = 0x1.1p0;

Apparently, the recent changes in the use of StringRef haven't been 
propagated to all places. Please find attached the trivial patch.

Cheers,
Enea Zaffanella.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: APFloat.patch
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20090820/6dde5d03/attachment.ksh>

On Thu, Aug 20, 2009 at 10:42 AM, Enea Zaffanella <zaffanella at
cs.unipr.it>wrote:
> When running clang in Debug mode, I am getting an assertion failure when
> parsing the following line:
>
> float ko = 0x1.1p0;
>
> Apparently, the recent changes in the use of StringRef haven't been
> propagated to all places. Please find attached the trivial patch.

Can you please also add a test to unittests/ADT/APFloatTest.cpp to validate
this, i.e., a test that would fail before this patch but passes with this
patch.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20090820/016062d4/attachment.html>

Fixed in r79450, along with an additional one of the same nature.

Thanks,
 - Daniel

On Thu, Aug 20, 2009 at 7:42 AM, Enea Zaffanella<zaffanella at
cs.unipr.it> wrote:
> Hello.
>
> When running clang in Debug mode, I am getting an assertion failure when
> parsing the following line:
>
> float ko = 0x1.1p0;
>
> Apparently, the recent changes in the use of StringRef haven't been
> propagated to all places. Please find attached the trivial patch.
>
> Cheers,
> Enea Zaffanella.
>
>
> Index: lib/Support/APFloat.cpp
> ==================================================================> ---
lib/Support/APFloat.cpp     (revision 79510)
> +++ lib/Support/APFloat.cpp     (working copy)
> @@ -2153,7 +2153,7 @@
>     integerPart hex_value;
>
>     if(*p == '.') {
> -      assert(dot == 0);
> +      assert(dot == s.end());
>       dot = p++;
>     }
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>

On Thu, Aug 20, 2009 at 10:13 AM, Daniel Dunbar<daniel at zuster.org>
wrote:
> Fixed in r79450, along with an additional one of the same nature.
>
> Thanks,
>  - Daniel
Thanks Daniel and Enea for finding this! I'm adding a lot more test
cases like this to APFloat's unittests so hopefully we won't have any
other bugs lying around.