[aspectj-dev] Fw: AspectJ bug status for 1.2rc1

[Adrian Colyer <adrian_colyer@xxxxxxxxxx>]

Forwarding Andy's summary of status
to the wider list - no reason for this discussion to be off-list that I
can see. 

My current proposal is that we do the
dual implementation fix for 59909 and then put out 1.2rc2  next week
(this kind of change will need some user validation in larger projects)
. Since a reasonable number of folk seem to be using the release candidate
via AJDT, we'll have to consider rev'ing AJDTat the same time too. My personal
plan is to keep focused on AJDT related development for the next couple
of days, and let Andy tackle 59909 on Tuesday of next week (Monday is a
public holiday in the UK).

-- Adrian
Adrian_Colyer@xxxxxxxxxx
----- Forwarded by Adrian
Colyer/UK/IBM on 28/04/2004 13:48 -----

Andrew Clement/UK/IBM

27/04/2004 17:28

To	Adrian Colyer/UK/IBM@IBMGB, aspectj@xxxxxxxxxxx, beatmik@xxxxxxx, eh-ajdev@xxxxxxxxxxxxx, George Harley1/UK/IBM@IBMGB, Julie Waterhouse/Toronto/IBM@IBMCA, wes@xxxxxxxxxxxxxx
cc
Subject	AspectJ bug status for 1.2rc1

Hi,

I'm about to disappear to Texas for
the rest of the week... so I wanted to share where I am at with the bugs.
 I've put the outstanding bugs into groups - the last group is the
most interesting as they are still 'up in the air'.  We also need
to decide if an RC2 is called for given the amount of fixes that seem to
be going in.

===
Bugs as of 27th April.  Not all
currently marked target 1.2, but these are the bugs that have come in since
1.2rc1 was released:

---

These are fixed for 1.2 already:

59596: Null pointer exception while
weaving java.awt.Label 
- Fixed.  Couldn't create testcase
though, until Wes raised 60015, part of which covers the same NPE.

59599: Race condition in test suite
for incremental 
- Believe its fixed by Adrian.

59895: NPE updating an aspect in incremental
mode 
- Fixed by Adrian I think.

---

These bugs are fixed enough for our
1.2 release, but the 'remainder' of each report either becomes a post 1.2
bug or an enhancement request.

58679: Regression from 1.1: NPE in CompilationResult
- Target should be changed to post 1.2.
 The temporary 'Integer.MAX_VALUE' fix is in right now which is good
enough for 1.2.

59397: NPE in compiler when using (an
unusual) declare warning a..
- Fix is in for the exception that occurs.
 Possibly needs moving to future enhancement now due to the issues
it raises about where join points exist for ITDd ctors/fields.

---

These can simply be deferred to post
1.2:

59636: -log option not producing output
- should it? 
- Agreed defer to post 1.2.

---

These are unreproduceable:

58567: Compiler weaves advice wrongly
if the core logic of a method is invoked within a return statement
- Still back with the raiser for more
info, I can't reproduce it.

59921: NoSuchMethodError: InstructionFactory.createNewArray
in e... 
- Back with raiser, I've proposed several
solutions.

---

These need discussion about
how to resolve them:

59208: Weaver fails in BCEL for large
classes
- We are blowing the limits for class
files with this one, we could put out an error message when this happens
rather than failing with a miserable exception.  But it is a bit of
an unusual scenario (using a program generator to create an unreasonable
program!).

59778: InterTypeMethodDeclaration.java:104

- I believe there is something here
we should fix, but I can't recreate it, there is no simple testcase so
far.  Code attached to the bug report is rather complicated.

59909: CFlowStack removesThreads to
late 
- Needs some kind of fix for 1.2 - Jim
proposes 3 options and suggests we do the 3rd one.  Will take a day
or two to implement and test.

60015: NPE, Incorrect XLint:unmatchedSuperTypeInCall
warning 
- NPE is same reason as in 59596.  But
still contains two issues to resolve:  I'm concerned about the message
that is produced being not quite right.  Wes highlights a problem
with the lint warning coming out at all.