I just took a look at various docs generated for

1. The new mismatch does not look good:

http://dtest.dlang.io/artifact/website-5eac54d55eb14787a4e4f8a8054e50f33fe164c1-d81ca42a781e6b8b64a57ee952c8a154/web/phobos-prerelease/std_algorithm_comparison.html#.mismatch

The ddox version is not better:

http://dtest.dlang.io/artifact/website-5eac54d55eb14787a4e4f8a8054e50f33fe164c1-d81ca42a781e6b8b64a57ee952c8a154/web/library-prerelease/std/algorithm/comparison/mismatch.html

Documentation should not mention `canon` etc but instead should expand the alias. This is an often-requested feature for ddoc. The template constraint looks bad too.

2. The public imports should link to the imported stuff:

http://dtest.dlang.io/artifact/website-5eac54d55eb14787a4e4f8a8054e50f33fe164c1-d81ca42a781e6b8b64a57ee952c8a154/web/phobos-prerelease/std_v2alpha_range_primitives.html

http://dtest.dlang.io/artifact/website-5eac54d55eb14787a4e4f8a8054e50f33fe164c1-d81ca42a781e6b8b64a57ee952c8a154/web/phobos-prerelease/std_v2alpha_meta.html

So probably at a minimum we should have the alias collapsing working for ddoc and/or ddox, or switch to adrdox. A solution would be beneficial regardless to the solution ultimately chosen for v2.

On Tuesday, 2 November 2021 at 22:07:19 UTC, Andrei Alexandrescu wrote:
> 2. The public imports should link to the imported stuff:

adrdox does this now on my box

http://dpldocs.info/experimental-docs/testimport.html

It lists out each selective import on its own line then you can override the doc comment if you want on the import itself, or otherwise it will look up the original item's comment.

This is what it looks like in a more real case right now:
http://dpldocs.info/experimental-docs/std.string.html#public-imports

Not sure I love it. I might take the selective ones and just merge them into the general members list. But undecided so far.

Aliases and selective imports work essentially the same way, so figuring it out for one will also do it for the other.

On Tuesday, 2 November 2021 at 22:07:19 UTC, Andrei Alexandrescu wrote:
> I just took a look at various docs generated for
>
> 1. The new mismatch does not look good:
>
> [...]

And this folks is why you want to run the experiment. Before it, you don't even know what questions will come up (in the hypothetical future), and it's quite hard answering a question you haven't even formulated yet 😎

On 2021-11-02 19:40, Imperatorn wrote:
> On Tuesday, 2 November 2021 at 22:07:19 UTC, Andrei Alexandrescu wrote:
>> I just took a look at various docs generated for
>>
>> 1. The new mismatch does not look good:
>>
>> [...]
> 
> And this folks is why you want to run the experiment. Before it, you don't even know what questions will come up (in the hypothetical future), and it's quite hard answering a question you haven't even formulated yet 😎

Great point, thanks.

Related update: Walter gave a cautious thumbs to the strategy and said we want to push exploration a bit further before taking any decisive steps.