[Jsr308-statements] Dealing with compiler optimizations

[Eugene Kuleshov]

  For what it worth, we can at least make a suggestion to disable 
optimizations and code elimination for annotated code fragments.

  But even then we should allow several ranges within the same 
annotation block, so it will be possible to say if those ranges actually 
belong to the same annotated block. This is also applies to the Local 
Variables variant of the reference_info structure for the main 
specification.

  regards,
  Eugene

PS: as far as I know JML is only working at the source-level, but BML 
(bytecode ML variant) may have something


Trevor Harmon wrote:
> In an earlier email [1], Mike made a good point about compiler 
> optimizations:
>
> "You need to discuss how you will handle compiler optimizations that 
> may reorder, duplicate, split, or move (parts of) statements. This is 
> one of the key technical challenges, in my view."
>
> This issue has been discussed peripherally on the main JSR-308 list, 
> but we should deal with it here directly. How should we address it? 
> What possible solutions are there?
>
> As we already know, several other tools (e.g., JML) already support 
> statement annotations via a proprietary mechanism. Does anyone know 
> how these tools handle compiler optimizations? Maybe we can borrow one 
> of their solutions for the proposal.
>
> Trevor
>
> [1] 
> https://lists.csail.mit.edu/pipermail/jsr308-statements/2007-March/000030.html 
>
>
>
> _______________________________________________
> Jsr308-statements mailing list
> Jsr308-statements at lists.csail.mit.edu
> https://lists.csail.mit.edu/mailman/listinfo/jsr308-statements