PING: PR 32095: Check for using symbols before they are typed

[Daniel Kraft]

Hi Tobias,

just in case this was lost somewhere...  But I'll be off the next week 
anyways, so don't hurry about this :)

Yours,
Daniel

Daniel Kraft wrote:
> Hi Tobias,
> 
> for your reference, attached is my current patch for PR 32095.  As it is 
> it handles the codes in the two related PRs as well as most of the 
> test-suite.  I did some minor changes to tests where I believe those are 
> obvious and ok (see in the patch), could you have a look at it if those 
> are really ok?
> 
> On the other hand, there remain three failing tests:
> 
> argument_checking_7.f90:  The compiler complains about the three 
> IMPLICIT declarations as the symbols used for the character-length are 
> all not yet typed.  This is a case that is disallowed unconditionally, 
> as it is the problem or PR 32095 (isn't it?).  What should I do here?  I 
> don't think this can be adapted easily and without changing much of the 
> test's semantics.
> 
> interface_12.f90:   The compiler does not like
>   character(f(x)) function test2 (x) result (r)
> as x is not yet typed in the character-definition.  Is this diagnostic 
> correct or do I have to handle some special cases?  If it is, can I 
> simply move the result-type specification after the imlicit statement? 
> But this seems to interfer with the test's intent.  Should the 
> result-type be interpreted in the context of the function's namespace? I 
> fear so.
> 
> result_in_spec_1.f90:  Seems to be quite the same as interface_12.f90.
> 
> Thanks,
> Daniel
> 


-- 
Done:     Arc-Bar-Sam-Val-Wiz, Dwa-Elf-Gno-Hum-Orc, Law-Neu-Cha, Fem-Mal
Underway: Cav-Dwa-Law-Fem
To go:    Cav-Hea-Kni-Mon-Pri-Ran-Rog-Tou