[m-rev.] for review: Fix compiler abort during liveness detection.

[Julien Fischer]

On Mon, 9 Sep 2013, Julien Fischer wrote:

>
> On Mon, 9 Sep 2013, Peter Wang wrote:
>
>> On Mon, 9 Sep 2013 13:43:28 +1000 (EST), Julien Fischer 
>> <jfischer at opturion.com> wrote:
>>> 
>>> Hi Peter,
>>> 
>>> On Wed, 4 Sep 2013, Peter Wang wrote:
>>> 
>>>> [I'm not certain of the fix]
>>> 
>>> Neither am I.
>>> 
>>> Do you know where the call to goal_info_get_post_births that is causing
>>> the aborts is?  In particular, is it whne detect_deadness (i.e. stage 2
>>> of of liveness analysis) is processing an if-then-else?
>> 
>> It happens immediately afterwards in the call to add_liveness_after_goal.
>
> So even sooner in fact.  I think you should go ahead and commit your
> fix since it's an improvement on the status quo.  I suggest you change
> the comment that you added to something like:
>
>    % XXX Initialize liveness-related fields, since some Other code in
>    % this module assumes that they are initialized.
>
> (There are at least two spots, the call to add_liveness_after_goal
> and also in detect_deadness_in_goal.  Presumably, another fix would be
> to replicate the check for condition that cannot fail at all of those
> points?)

I meant to say "the condition that cannot succeed ... " in that
last sentence.)

Cheers,
Julien.