[cfe-commits] r64627 - /cfe/trunk/lib/Analysis/RangeConstraintManager.cpp

[Ted Kremenek]

On Feb 16, 2009, at 8:51 AM, Ben Laurie wrote:

> On Mon, Feb 16, 2009 at 4:48 PM, Ben Laurie <benl at google.com> wrote:
>> On Mon, Feb 16, 2009 at 4:26 PM, Ted Kremenek <kremenek at apple.com>  
>> wrote:
>>> On Feb 16, 2009, at 1:19 AM, Ben Laurie <benl at google.com> wrote:
>>>
>>>> On Mon, Feb 16, 2009 at 4:54 AM, Ted Kremenek  
>>>> <kremenek at apple.com> wrote:
>>>>>
>>>>> Author: kremenek
>>>>> Date: Sun Feb 15 22:54:20 2009
>>>>> New Revision: 64627
>>>>>
>>>>> URL: http://llvm.org/viewvc/llvm-project?rev=64627&view=rev
>>>>> Log:
>>>>> Do not register 'RangeConstraintManager' as the default
>>>>> ConstraintManager.
>>>>
>>>> I should note that this didn't work, by the way.
>>>
>>> Hi Ben,
>>>
>>> I'm more than happy to fix my mistakes, but your comment doesn't  
>>> contain any
>>> actionable information.  Can you clarify what you mean by "this  
>>> didn't
>>> work"?. That could mean a variety of issues.  Are you seeing that
>>> RangeConstraintManager is not being used at all now (despite  
>>> whatever
>>> options you pass to the command line)?  If that is the case,  I  
>>> suspect the
>>> analagous line in BasicConstraintManager.c needs to be removed as  
>>> well (I
>>> can't tell right now if there is one since I'm not near an editor).
>>
>> What i meant was that adding
>>
>> RegisterConstraintManager X(CreateRangeConstraintManager);
>>
>> did not cause the RangeConstraintManager to be used when there was no
>> command-line option to enable it. I did not investigate further since
>> this is clearly the wrong way to do it.
>
> BTW, it seemed to me that this mechanism would be more useful if it
> registered the constraint manager and some string identifying it, and
> then used a command-line option to select one of the registered
> managers...

I agree, although the mechanism has to be general enough that the  
individual ConstraintManagers don't assume that they are being used  
from a command-line driver.  The logic you mention would mostly go  
into AnalysisConsumer (which is driver specific).  Feel free to take a  
stab at it if you are interested.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20090216/3d8e16ca/attachment.html>