[Laszlo-dev] For Review: Change 20080724-dda-2 Summary: minor cleanup of SWF6 compiler transforms

[Donald Anderson]

Thanks.  I also found this very good page:

http://www.m2osw.com/fr/swf_alexref.html

On Jul 24, 2008, at 7:00 PM, P T Withington wrote:

> Approved.  Rats on the no new instructions!
>
> Possibly worth investigating for more info on swf8 byte codes: http://www.mtasc.org/
>
> On 2008-07-24, at 17:50EDT, Donald Anderson wrote:
>
>> Change 20080724-dda-2 by dda at lester.local on 2008-07-24 16:43:04 EDT
>>   in /Users/dda/laszlo/src/svn/openlaszlo/trunk-a
>>   for http://svn.openlaszlo.org/openlaszlo/trunk
>>
>> Summary: minor cleanup of SWF6 compiler transforms
>>
>> New Features:
>>
>> Bugs Fixed: LPP-6669 (Remove swf6 byte-code kludges from  
>> CodeGenerator.java)
>>
>> Technical Reviewer: ptw (pending)
>> QA Reviewer: (pending)
>> Doc Reviewer: (pending)
>>
>> Documentation:
>>
>> Release Notes:
>>
>> Details:
>> Most of the issues implied by the LPP-6669 cannot be fixed - but  
>> details are listed here.
>>
>> 0)  Removed some code specific to swf6 involving using lower case  
>> for variable and register names
>>
>> Examined and tested the various transforms in question:
>>
>> 1)
>>   // swf6 returns undefined for comparisons with NaN, it
>>   // is supposed to return false (note that you cannot
>>   // eliminate one NOT by inverting the sense of the
>>   // comparison
>>   BinopInstrs.put(ParserConstants.LT, new Instruction[]  
>> {Instructions.LessThan, Instructions.NOT, Instructions.NOT});
>>
>> Tried setting this to just Instructions.LesThan, but smokecheck  
>> gives errors on NaN comparisons.
>>
>> 2)
>>   // swf6 does not have GE or LE, but inverting the
>>   // complement operator does not work for NaN ordering
>>   // Luckily, LogicalOr coerces undefined to false, so we
>>   // don't have to play the NOT NOT trick above
>>   BinopInstrs.put(ParserConstants.LE, new Instruction[]
>>
>> Tried setting LE to { Instructions.GT, Instructions.NOT }
>>  smokecheck also gives errors on NaN comparisons.
>> For kicks, also tried setting LE to { Instructions.GT,  
>> Instructions.NOT , Instructions.NOT , Instructions.NOT }
>>  to see if the 'double NOT' trick might work, but also gives the  
>> same smokecheck errors.
>>
>> Also examined flasm source to see if GE or LE has been added (they  
>> have not).
>>
>> 3)
>>   // swf6 does not have NE or SNE either, but inverting
>>   // the complement is correct for NaN
>>
>> There are no new opcodes for NE or SNE (at least as known to flasm)
>>
>> 4)
>>   // Approximate a in b as b.a =! void 0
>>
>> No new opcodes for 'in'
>>
>> 5)  Special case of 'is' in translateBinaryExpression.
>>
>> No new opcodes for 'is'
>>
>> 6)  Special case of 'cast' in translateBinaryExpression.
>>
>> There is a SWF 'CAST' operator that leaves a 'null' on the stack in  
>> case of error.
>> This is not being utilized, and there's already a separate TODO for  
>> this item.
>> It didn't seem to be under the purvue of this Jira bug.
>>
>>
>> Tests:
>>  regression: smokecheck swf8/dhtml
>>
>> Files:
>> M      WEB-INF/lps/server/src/org/openlaszlo/sc/CodeGenerator.java
>>
>> Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20080724-dda-2.tar
>>
>>
>>
>> --
>>
>> Don Anderson
>> Java/C/C++, Berkeley DB, systems consultant
>>
>> voice: 617-547-7881
>> email: dda at ddanderson.com
>> www: http://www.ddanderson.com
>>
>>
>>
>


--

Don Anderson
Java/C/C++, Berkeley DB, systems consultant

voice: 617-547-7881
email: dda at ddanderson.com
www: http://www.ddanderson.com



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.openlaszlo.org/pipermail/laszlo-dev/attachments/20080725/fa9f1c9f/attachment.html