C++: Namespace name dropped in XML documentation for base classes (Origin: bugzilla #765812)

[doxygen]

status NEW severity major in component general for ---
Reported in version 1.8.11 on platform Other
Assigned to: Dimitri van Heesch

On 2016-04-29 15:28:56 +0000, DT wrote:

===== Sample source code: =====

struct a1 : public d::c
{
};

struct a2 : public d::c
{
};

struct a3 : public d::c
{
};


===================

When generating XML documentation for this code, only the first structure
will have full namespace name documented for base class:

    <inheritancegraph>
      <node id="0">
        <label>a1</label>
        <link refid="structa1"/>
        <childnode refid="1" relation="public-inheritance">
        </childnode>
      </node>
      <node id="1">
        <label>d::c</label> <<<---- Notice this CORRECT line: d::c
      </node>
    </inheritancegraph>

But for all other structs in the file namespace name will be dropped for
base class:

    <inheritancegraph>
      <node id="5">
        <label>c</label> <<<---- Notice this WRONG line: "c", namespace "d"
dropped!
      </node>
      <node id="4">
        <label>a2</label>
        <link refid="structa2"/>
        <childnode refid="5" relation="public-inheritance">
        </childnode>
      </node>
    </inheritancegraph>

[excitoon]

FYI. This bug was introduced between Release_1_7_1 and Release_1_7_2.

[albert-github]

The question here is should the relation be to "d" or to "d::c" but anyway it should be consistent.
To reproduce the bug it is important to set HIDE_UNDOC_RELATIONS = NO, besides setting the EXTRACT_.... = YES.
When looking to the code in doxygen.cpp function findClassRelation here are 2 calls to insertBaseClass the second one is called for a1 and the first one for a2. The relevant argument here is the 2nd argument which is filled differently.

I'm not sure what the reason of the different handling is and what the consequences would be when changing either one of them.

(Problem is also shown in inheritance and collaboration graphs in e.g. HTML)

[excitoon]

More information is better that less, I guess. We depend on this behavior and expect full name of a class (otherwise, how would one guess which one is mentioned?)

[excitoon]

This is a change which introduced a bug: f0e4767 .

[excitoon]

Exactly this line.
f0e4767#diff-1c8be53f930af9a4ffb709d1e81bc430R3641
Can someone explain what is going on?
@albert-github @doxygen
#3841 #5199

[XonqNopp]

Any chance to see this fixed?

[excitoon]

It *is* fixed, in my PR 😎

…

On Fri, Jul 24, 2020, 7:58 AM XonqNopp ***@***.***> wrote: Any chance to see this fixed? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5993 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AARGSB4ZA4SWKIFXVK7NMATR5EIGRANCNFSM4FH4ZESQ> .