[modules-discuss] A paper about JAM

[Stephen J. McConnell]

On shallow validation (section 3) ...

This section addresses the pros and cons of depth-first validation (self,
then imports) versus reverse depth-first fashion (imports, then self).  The
example in section 6.4 focuses on a scenario involving multiple duplicate
fully qualified names (which according to 277 draft is an to avoided).
However, the sources for the modules package includes a ModuleDefinition
class which includes the AllowShadowing annotation which in effect moves us
into a gray area (discussion of and/or mechanisms to allow shadowing are not
discussed in the 277 draft but only appear in the sources at super-package
level).  From this rather loose platform the paper is asserting the
following statement at the end of section 3:

  Therefore, the only guarantee validation gives is that, if it
  succeeds, there will be no unexpected behaviour due to the
  unnatural class lookup function. 

I think that some additional clarity is needed here - but I also think that
depth-first versus reverse depth-first is a subject that has not been
discussed to any significant extent (at least I have not seen this on the
294 or 277 lists).  I think this subject is worthy of additional discussion
(as I for one have some opinions critical of the scoping of AllowShadowing
annotation).

In effect - within section 3 the report makes a distinction between
depth-first versus reverse depth-first (which is an opinion raised by the
paper as opposed to an opinions put forward in the 277 draft - but all the
same is an interesting opinion).  The paper subsequently draws on this
opinion in terms of the emphasis assigned to conclusions which appear to
qualify 277 relative to concepts established in the paper.  I.e. the modules
stuff is in a gray area on this subject and the paper is asserting points of
view but basing these points of view on its own assertions - and things get
confused.

Cheers, Steve.



> -----Original Message-----
> From: modules-discuss-bounces at openjdk.java.net 
> [mailto:modules-discuss-bounces at openjdk.java.net] On Behalf 
> Of Stephen J. McConnell
> Sent: Friday, 6 July 2007 3:02 PM
> To: modules-discuss at openjdk.java.net
> Subject: Re: [modules-discuss] A paper about JAM
> 
>  
> 
> Rok Strnisa wrote:
> > We would be grateful for any
> > comments/suggestions before the OOPSLA camera-ready 
> deadline on July 
> > 30, 2007.
> 
> I'm still working my way though the document but I hit the 
> following statement at the end of section 3.6:
> 
>   "JSR-277 states that repositories will be implemented
>   with classloaders."
> 
> That doesn't seem correct to me.
> 
> In the meantime I'll continue reading.
> 
> Cheers, Steve.
> 
> _______________________________________________
> modules-discuss mailing list
> modules-discuss at openjdk.java.net
> http://mail.openjdk.java.net/mailman/listinfo/modules-discuss