• =?UTF-8?B?Q2hyaXN0aWFuIFLDtnNzZWw=?= <christian.roessel gmx.de> Aug 08 2008
• Matthew <matthewwilson acm.org> Aug 09 2008
• Matt Wilson <matthewwilson acm.org> Aug 11 2008

=?UTF-8?B?Q2hyaXN0aWFuIFLDtnNzZWw=?= <christian.roessel gmx.de> writes:

Hi Matthew,

the copy-ctor accesses rhs[0] in

    pod_copy_n(begin_(), &rhs[0], size());

If rhs is empty, we get a "Requested index is out of range" assertion. A
possible fix is:

    if (! rhs.empty()) {
      pod_copy_n(begin_(), &rhs[0], size());
    }

Is this workaround sufficcient or do I miss something?
I found this bug in 1.9.42 and 1.9.45.

Best regards,
Christian

Matthew <matthewwilson acm.org> writes:

Hi Christian

I'm working on this right now. The fix is to change line 489 to 

    pod_copy_n(begin_(), rhs.begin_(), size());

This has caused me to finally start the release of the new (1.10+) STLSoft
unit-testing facility, which you should see available on the site sometime in
the next day or so.

Thanks

Matt

Christian Rössel Wrote:

 Hi Matthew,
 
 the copy-ctor accesses rhs[0] in
 
     pod_copy_n(begin_(), &rhs[0], size());
 
 If rhs is empty, we get a "Requested index is out of range" assertion. A
 possible fix is:
 
     if (! rhs.empty()) {
       pod_copy_n(begin_(), &rhs[0], size());
     }
 
 Is this workaround sufficcient or do I miss something?
 I found this bug in 1.9.42 and 1.9.45.
 
 Best regards,
 Christian

Matt Wilson <matthewwilson acm.org> writes:

The release of 1.9.46 takes into account this defect (and several others).

Cheers

Matt

Christian Rössel Wrote:

 Hi Matthew,
 
 the copy-ctor accesses rhs[0] in
 
     pod_copy_n(begin_(), &rhs[0], size());
 
 If rhs is empty, we get a "Requested index is out of range" assertion. A
 possible fix is:
 
     if (! rhs.empty()) {
       pod_copy_n(begin_(), &rhs[0], size());
     }
 
 Is this workaround sufficcient or do I miss something?
 I found this bug in 1.9.42 and 1.9.45.
 
 Best regards,
 Christian