Yay!

On 2010-07-01, at 16:22, Henry Minsky wrote:

> ok I was setting it in the wrong place. Now it's properly evaluating the 'if
> ($debug)' at compile time.
> 
> On Thu, Jul 1, 2010 at 4:16 PM, P T Withington <[email protected]>wrote:
> 
>> Look at Compiler where it clones the props and adds the compile-time
>> constants before it creates the object writer.  You need to do the same.
>> IWBNI there were a 'one true way' to do that...
>> 
>> On 2010-07-01, at 16:15, Henry Minsky wrote:
>> 
>>> Maybe I'm setting them on the wrong properties table, lemme check.
>>> 
>>> On Thu, Jul 1, 2010 at 4:06 PM, P T Withington <[email protected]
>>> wrote:
>>> 
>>>> On 2010-07-01, at 15:47, Henry Minsky wrote:
>>>> 
>>>>>> 
>>>>>> Are you looking at the output of the tag compiler, or the output of
>> the
>>>>>> script compler?
>>>>>> 
>>>>> 
>>>>> I'm looking at the as3 output from the SWF9 backend of the script
>>>> compiler,
>>>>> and I still
>>>>> see the "if ($debug)" (as andre pointed out).  Is something being
>>>> bypassed
>>>>> for
>>>>> constraint calculations?
>>>> 
>>>> Hm...  That should not be the case.
>>>> 
>>>> In http://jira.openlaszlo.org/jira/browse/LPP-9017 we backed out the
>> stub
>>>> for `$lzc$validateReferenceDependencies` that was (incorrectly) added in
>>>> r16402 for http://jira.openlaszlo.org/jira/browse/LPP-9003.  After
>> backing
>>>> that out, we don't get errors in DHTML or as3 compiles (which we would,
>> if
>>>> the optimization were not happening).
>>>> 
>>>> This makes me think that the import compiler is doing something
>> different
>>>> that is preventing the optimization there.  The compile-time constants
>> have
>>>> to be set in the object writer, not just the tag compiler.
>>>> 
>>>> 
>>> 
>>> 
>>> --
>>> Henry Minsky
>>> Software Architect
>>> [email protected]
>> 
>> 
> 
> 
> -- 
> Henry Minsky
> Software Architect
> [email protected]


Reply via email to