Re: [PATCH] Logical short circuiting with Fortran argument lists

[Simon Marchi <simon.marchi@polymtl.ca>]

On 2018-08-06 12:35, Richard Bunt wrote:
> Hi Simon,
> 
> Many thanks for the review.
> 
> To clarify, do you observe the test passing with no changes to eval.c 
> or
> just when the additions are applied?

I mean that I saw the test passing only with the first hunk applied (the 
one that adds code).

> If the former is the case, may I ask which compiler are you using? I
> have retested with 953473375500 and I see 13 tests fail without the
> additions. If it's the latter, the deletions are not strictly needed 
> but
> my analysis determined that this code was unreachable after this patch.
> 
> My analysis consisted of checking for regressions in the test suite (of
> which there were none) and examining all uses of noside from the new
> early termination to eval_call in TYPE_CODE_FUNC. There are no
> assignments and it's passed to all functions by value.

Ah indeed, I read it wrong the first time.  I thought that the second 
hunk was in a totally different case of the main switch (handling of a 
different OP_*.  But now I see that this is all under the 
"OP_F77_UNDETERMINED_ARGLIST" case, and that we'll never reach the 
bottom part with "noside == EVAL_SKIP".

It LGTM then.

Simon