pr28974 Extremely slow compilation of enumerated DATA statements.

[Jerry DeLisle]

Bud Davis wrote:
> below is a patch that addresses the subject PR. it
> works well with contrived sample programs, for what
> that is worth :)
> 
> it has some stylistic problems; some hunks of code
> need to be segregated out into procedures; some types
> could be dealt with cleaner; and the logic needs to be
> optimised a bit....but it does deal with the quadratic
> behavour.
> 
> request that those interested try this patch out and
> look for problems with real code.  i am curious as to
> any adverse impacts this patch has on memory required
> for compilation and or compilation speed. 
> 
> i don't think this fix is essential for 4.2, and i
> plan on submitting a polished patch when the 4.3
> branch opens up.
> 
> --bud
The patch applies cleanly, regression tested fine on i686-linux.

I also tested with NIST and LAPACK, no regressions.

So far so good.

Jerry