Re: [RFC] c++: Print function template parms when relevant (was: [PATCH v4] c++: Add gnu::diagnose_as attribute)

[Jason Merrill <jason@redhat.com>]

On 11/17/21 04:04, Matthias Kretz wrote:
> On Wednesday, 17 November 2021 07:09:18 CET Jason Merrill wrote:
>>> -  if (CHECKING_P)
>>> -    SET_NON_DEFAULT_TEMPLATE_ARGS_COUNT (a, TREE_VEC_LENGTH (a));
>>> +  SET_NON_DEFAULT_TEMPLATE_ARGS_COUNT (a, nondefault);
>>
>> should have been
>>
>> if (CHECKING_P || nondefault != TREE_VEC_LENGTH (a))
>>     SET_NON_DEFAULT_TEMPLATE_ARGS_COUNT (a, nondefault);
> 
> TBH, I don't understand the purpose of CHECKING_P here, or rather it makes me
> nervous because AFAIU I'm only testing with CHECKING_P enabled. Why make
> behavior dependent on CHECKING_P? I expected CHECKING_P to basically only add
> more assertions.

The idea when NON_DEFAULT_TEMPLATE_ARGS_COUNT was added years back was 
to leave the TREE_CHAIN null when !CHECKING_P and treat that as 
equivalent to TREE_VEC_LENGTH (args).  But perhaps you're right that 
it's not a savings worth the complexity.

>>> (copy_template_args): Jason?
>>
>> Only copy the non-default template args count on TREE_VECs that should
>> have it.
> 
> Why not simply set the count on all args? Is it a performance concern? The
> INTEGER_CST the TREE_CHAIN has to point to exists anyway, so it's not wasting
> any memory, right?

In this case the TREE_VEC we're excluding is the one wrapping multiple 
levels of template args; it doesn't contain args directly, so setting 
NON_DEFAULT_ARGS_COUNT on it doesn't make sense.

>>> +  /* Pretty print only template instantiations. Don't pretty print
>>> explicit
>>> +     specializations like 'template <> void fun<int> (int)'.
>>
>> This seems like a significant change of behavior unrelated to printing
>> default template arguments.  What's the rationale for handling
>> specializations differently from instantiations?
> 
> Right, this is about "The general idea of this change is to print template
> parms wherever they would appear in the source code as well".
> 
> Initially, the change to print function template arguments/parameters only if
> the args were explicitly specified lead to printing 'void fun (T) [with T =
> ...]' or 'template <> void fun (int)'. Both are not telling the full story,
> even if the former is how the function would be called.

and the latter is how I expect the specialization to be declared, not 
with the deducible template argument made explicit.

> But if the reader
> should quickly recognize what code is getting called, it is helpful to see
> right away that a function template specialization is called. (It might also
> reveal an implementation detail of a library, so it's not 100% obvious how to
> choose here.) Also, saying 'T = int' is kind of wrong. Yes, 'int' was deduced.
> But there's no T in fun<int>:
> 
> template <class T> void fun (T);
> template <> void fun<int> (int);

There's a T in the template, and as you said above, that's how it's 
called (and mangled).

> __FUNCTION__ was 'fun<int>' all the time, but __PRETTY_FUNCTION__ was 'void
> fun(T) [with T = int]'.

Isn't that true for instantiations, as well?

> It's more consistent that __PRETTY_FUNCTION__ contains __FUNCTION__, IMHO

I suppose, but I don't see that as a strong enough motivation to mix 
this up.

> so it would have to be at least 'void fun<int>(T) [with T
> = int]'. But that's strange: How it uses T and int for the same type. So I
> settled on 'void fun<int>(int)'.
> 
>> I also don't understand the purpose of TFF_AS_PRIMARY.
> 
> dump_function_decl generalizes the TEMPLATE_DECL (if flag_pretty_templates is
> true) and, before this change, passes the generalized TEMPLATE_DECL to
> dump_type (... DECL_CONTEXT (t) ...) and dump_function_name (... t ...).
> That's why the whole template is printed as primary template (i.e. with
> template parms instead of template args, as is needed for
> flag_pretty_templates). But this drops the count of non-default template args.

Ah, you're trying to omit defaulted parms from the <list>?  I'm not sure 
that's necessary, leaving them out of the [with ...] list should be 
sufficient.

> To retain the count, dump_type and dump_function_name need to be called with
> the original TEMPLATE_DECL. But if I do this, pretty-templates is broken.
> 'template <class T> struct A { template <class U> void f(T, U); };' would
> print as 'A<int>::f<float>(T, U) [with U = float, T = int]'. To get back to
> 'A<T>::f<U>(T, U) [with U = float, T = int]' I needed to tell
> dump_template_parms that even though the template args are there, it should
> print only the template parms. The most obvious way to do that was to carry it
> through via flags.
> 
> Note that this creates another problem. Given
> 
> template <class T0, class T1 = int> struct Outer {
>    template <class T, class U> struct A;
>    template <class X> struct A<X, int> {
>      void f();
>    };
> };
> 
> we want to print e.g. 'void Outer<T0>::A<X, int>::f() [with X = int, T0 =
> int]', but certainly not 'void Outer<T0>::A<T, U>::f() [with X = int, T0 =
> int]'. However, specialized_t holds A<int, int> which is printed as A<T, U>
> with TFF_AS_PRIMARY. Only most_general_template of the function's
> TEMPLATE_DECL can give us A<X, int> as DECL_CONTEXT.
> 
> I have a solution in the diagnose_as patch, where I had to solve a similar
> problem because for the diagnose_as attribute (dump_template_scope).
> 
>>> +/* Print function template parameters if:
>>> +   1. t is template, and
>>> +   2. flags did not request "show only template-name", and
>>> +   3. t is a specialization of a template (Why is this needed? This was
>>> present +      since 1999 via !DECL_FRIEND_PSEUDO_TEMPLATE_INSTANTIATION:
>>> "Don't crash if +      given a friend pseudo-instantiation". The
>>> DECL_USE_TEMPLATE should likely +      inform the PRIMARY parameter of
>>> dump_template_parms.), and
>>
>> Good question.  It seems that the existing
>> !DECL_FRIEND_PSEUDO_TEMPLATE_INSTANTIATION has mostly been excluding the
>> most general template; removing that line changes things like
>>
>> int bar(T) [with T = int]
>>
>> to
>>
>> int bar<T>(T) [with T = int]
>>
>>
>>
>>> +   4. either
>>> +      - flags requests to show no function arguments, in which case
>>> deduced +        types could be hidden, or
>>> +      - at least one function template argument was given explicitly, or
>>> +      - we're printing a DWARF name,
>>
>> ...but perhaps this can replace the above.
> 
> I'll rerun the tests with DECL_USE_TEMPLATE moved to the PRIMARY parameter of
> dump_template_parms.
> 
>> Though, why do we want
>> different behavior here when printing a DWARF name?
> 
> Sorry, I should have asked ... I also added the same issue as an open question
> on the diagnose_as patch. When I ran the whole testsuite I had failures in the
> DWARF tests. This change resolved them. I don't know enough about how those
> strings are used and whether they may change between GCC versions. Anyway, the
> DWARF strings in that test had only the function name and template argument
> list (i.e. no function arguments).

If they only have arguments, isn't TFF_NO_FUNCTION_ARGUMENTS set?

> If the template argument list was removed
> (as was the case without the condition here), the DWARF strings lost important
> information, giving several different functions the same name. That seemed
> like a regression. So either the DWARF strings need to include the function
> arguments, or we need this condition to keep showing the template arguments
> independent of whether they were explicitly given or not.
> 
>>> +   5. either
>>> +      - t is a member friend template of a template class (see
>>> DECL_TI_TEMPLATE +        documentation), or
>>> +      -
>>
>> Missing the last item.  :)
> 
> Oh, I got distracted while trying to figure this out. :) I'll try to
> understand why PRIMARY_TEMPLATE_P (DECL_TI_TEMPLATE (t)) is needed and
> document it.
>