554617 - getlex opcode cannot be assembled without a multiname qualifier

Status: RESOLVED WONTFIX

(Tamarin Graveyard :: Tools, defect)

Description

[simonzack]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2.2) Gecko/20100316 Firefox/3.6.2 (.NET CLR 3.5.30729)
Build Identifier: 

assembling the opcode:
    57      getlex        	flash.events::MouseEvent

(which is what abcdump generated)
gives a syntax error


Reproducible: Always

Steps to Reproduce:
1.assemble the above line
2.
3.
Actual Results:  
syntax error

Expected Results:  
assembles correctly even without multiname qualifiers

Comment 1

[simonzack]

Digging a little deeper, this is actually a grammar bug

the rule: ".exception"
and the rule: namespace_specification conflict, and some others do as well (all the ones that begin with DOT)

this is the actual bug

Comment 2

[Edwin Smith]

I've assigned the bug to the abcasm author, but if you have a suggested fix, feel free to post it with or without a test case.  Thanks!

Comment 3

[Tom Harwood]

abcasm name processing in general is still alpha-level at best.  Thanks for the analysis!

Comment 4

[simonzack]

Attachment: grammar fix

Comment 5

[simonzack]

no probs
attached a fix for grammar
btw I've only fixed the existing bug in multi-name processing
now "{flash.event}" won't conflict with ".exception"
but there is still no way to write unqualified names
atm abcdump just outputs ns1::foo
Is abcasm in alpha stage too? Most of the abs files I feed it aren't processed at all

Comment 6

[Tom Harwood]

(In reply to comment #5)

> Is abcasm in alpha stage too? Most of the abs files I feed it aren't processed
> at all

Yes, the assembler is an Alpha stage project.  One of its goals is to round-trip files (i.e., abcdump an ABC, assemble the dump output, get an equivalent ABC back), but that's a long-term goal.

Comment 7

[Tom Harwood]

Attachment: grammar fix as a patch

Resubmitting previous attachment as a patch.

Comment 8

[Tom Harwood]

The patch as written is giving lots of errors in regression testing, but I think syntactic predicates will do the trick.