Re: [wtp-releng] Compile errors in org.eclipse.jst.server.generic.core?

[David M Williams <david_williams@xxxxxxxxxx>]

I do not have any objections to moving
to 1.5 for this bundle ... if that is the correct thing to do. But it seems
that we should call this to the attention of EMF and/or JDT compile team
and have them confirm there is no bug or issue that would effect others
in perhaps even worse ways?  

You've done so much already, Larry,
but I suggest you open a bug ... I guess first in EMF ... and document
this issue and your findings. I would not be so concerned except it is
odd that simply moving to a new compiler version produces the error. I
guess its hard to tell if the compiler was wrong before, and is now working
correctly or if it is incorrect now. 

At the moment, though, I'm going to
assume this is not an issue for our M4 deliverable. If anyone wants to
test with final version of M4 (and M4a) that'd be great, but seems at most
it would be an issue only with 
org.eclipse.jst.server.generic.core
... and ... what would that effect? Seems more a "major" bug
than a blocker (if there is a runtime error). Correct me if I'm wrong.


Thanks so much. 





From:      
 Larry Isaacs <Larry.Isaacs@xxxxxxx>
To:      
 Webtools releng discussion
list <wtp-releng@xxxxxxxxxxx>
Date:      
 12/15/2010 04:28 PM
Subject:    
   Re: [wtp-releng]
Compile errors in org.eclipse.jst.server.generic.core?
Sent by:    
   wtp-releng-bounces@xxxxxxxxxxx

---

I think I re-generated the
model appropriately, creating new classes in a package that didn’t contain
the “internal”.  These re-generated classes have the same compile
errors that the committed ones do.  However, if I increase the compiler
compliance level to “1.5” from “1.4”, these compile errors go away.
 I get numerous warnings, but adding “<?>” to the raw types
seems to fix these.
 
Larry 
 
From: Larry Isaacs 
Sent: Wednesday, December 15, 2010 3:14 PM
To: 'Webtools releng discussion list'
Subject: RE: [wtp-releng] Compile errors in org.eclipse.jst.server.generic.core?
 
Not surprisingly 3.4a didn’t
change the result.  With the EMF interfaces and classes I’ve peeked
at so far, I’m having more trouble figuring out how the code compiles
with 3.6 than how it doesn’t with 3.7.  They may have fixed a compiler
bug or implemented some clarification about what is legal for interface
casting.  We had a similar kind of issue moving some SAS code from
the 3.4 compiler to the 3.6 compiler.  I’ll see if I can manage to
re-generate the classes and check if anything changes in the generated
code, in case something changed in EMF that mandates this.
 
Larry
 
From: wtp-releng-bounces@xxxxxxxxxxx
[mailto:wtp-releng-bounces@xxxxxxxxxxx]
On Behalf Of David M Williams
Sent: Wednesday, December 15, 2010 2:19 PM
To: Webtools releng discussion list
Subject: Re: [wtp-releng] Compile errors in org.eclipse.jst.server.generic.core?
 
Thanks Larry for your quick investigation
and letting me know you could reproduce. 

I have (just) discovered that EMF has (just) provided an M4a build, due
to some "packaging" errors. So .... there is a possibility that
it is related to that ... though seems unlikely. See 
https://bugs.eclipse.org/bugs/show_bug.cgi?id=332474
https://bugs.eclipse.org/bugs/show_bug.cgi?id=332490

So ... first thing I'll try is switching to that M4a version of EMF ...
if we still get the compile errors there I will open a bug. 

Thanks again, 






From:        Larry
Isaacs <Larry.Isaacs@xxxxxxx>
To:        Webtools
releng discussion list <wtp-releng@xxxxxxxxxxx>
Date:        12/15/2010
01:46 PM 
Subject:        Re:
[wtp-releng] Compile errors in org.eclipse.jst.server.generic.core?
Sent by:        wtp-releng-bounces@xxxxxxxxxxx

---

David, 
  
I can confirm that this is a 3.7 compiler issue related somehow to the
extending/implementing of interfaces.  The same code compiles fine
in my 3.6 development environment and duplicates the compile errors in
the 3.7 environment I just set up.  Unfortunately, trying to understand
how the EMF interfaces/classes are pieced together has revealed how little
I understand EMF at the moment.  The problem classes appear to be
EMF generated, so with my very limited knowledge of EMF, I would not be
able to supply a timely fix, other than to switch back to the 3.6 compiler.
Larry 
  
From: wtp-releng-bounces@xxxxxxxxxxx
[mailto:wtp-releng-bounces@xxxxxxxxxxx]
On Behalf Of David M Williams
Sent: Wednesday, December 15, 2010 4:45 AM
To: wtp-releng@xxxxxxxxxxx
Subject: [wtp-releng] Compile errors in org.eclipse.jst.server.generic.core?
  
In the latest 3.3 build, there were a number of "illegal cast"
compile errors in org.eclipse.jst.server.generic.core. 

Was this due to a change in org.eclipse.jst.server.generic.core? 

I ask because I am fairly sure all the pre-reqs were exactly the same as
previous build (where there were no compile errors) but I did change the
version of our "base builder" which (mostly) means previously
builds were compiled with a 3.6 based JDT compiler, but now are compiled
with a 3.7 M4 based compiler. Could there have been a change there to cause
this? Or a change in defaults ... former warnings now caught as errors?
  

Even though we do not have a "declared build" this week, I'd
appreciate understanding this issue quickly (EOD Wednesday?), to know if
we have a problem with our M4 deliverables ... or, if someone just casually
checked in some code with errors? 

Thanks, 
_______________________________________________
wtp-releng mailing list
wtp-releng@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/wtp-releng_______________________________________________
wtp-releng mailing list
wtp-releng@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/wtp-releng