Re: [j-devel] Progress on special variable binding fixes

[XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]

And ofcourss, as soon as you ask for help, you find the solution. Fix committed.

On Sat, Sep 13, 2008 at 4:09 PM, XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX wrote:
> Well, today I fixed another 2 tests which were special variable
> related. This brings us down to 56 failures.
>
> Unfortunately, some of the fixes of last week introduced new failures:
> FLET.40 and LABLES.43. Fortunately, these were also special-binding
> related *and* the problem they detect is probably caused by the one
> and the same source.
>
> So, we're progressing nicely, but some help in finding this newly
> uncovered source of breakage would be great!
>
>
> Some remarks regarding the Java source code so far:
>
>  - I found there to be quite a bit of code-duplication. (I've started
> to resolve some of that by merging some code paths which differ only
> by 1 or 2 statements)
>  - The code is largely undocumented (this doesn't really help when
> actually trying to find out the meaning of the code)
>  - The use of the "finally" statement in order to restore old special
> bindings on function exit is highly under-used: in many places,
> variables and return statements are tracked "by hand", without using
> "finally"; more "finally" could considerably clean up the code.
>  - try {} finally {} catch {} blocks seem to have been kept as small
> as possible, having lots of code outside the scope of
> special-variable-state restoration.
>
>
> Well, anyway, as I said, please help clearing up the FLET.40 mistery.
>
>
> Bye,
>
> Erik.
>