Redesign of dlang.org

[Andrei Alexandrescu via Digitalmars-d]

On 12/24/15 8:26 AM, Thomas Mader wrote:
> On Thursday, 24 December 2015 at 06:43:32 UTC, Andrei Alexandrescu wrote:
>> Currently dlang.org has over 62KLOC of Ddoc source, so any significant
>> surgery on it will be a large effort. Dropping ddoc means we'd need to
>> use another templating engine (getting back to raw html would be too
>> much trouble), and 10 people have 11 ideas about which template engine
>> is used by "everyone".
>>
>> I can give you right now an estimate - dropping ddoc and replacing it
>> with vibe.d is unlikely to be a landslide success. When the alternate
>> documentation was introduced using vibe.d, my hope was that everybody
>> would be all over it like a cheap white suit on rice, and that the use
>> of vibe.d would organically grow to make the stdlib documentation
>> stellar, and then engulf the main site. Sadly participation was scant,
>> and we had a couple of vibe.d-related situations in which the
>> maintainer division (ahem... Vladimir and myself) had no idea on what
>> to do and had nobody to rely on.
>
> Thanks for those details your decision is much more clear now for me.
> I didn't know that the documentation is switched to vibe.d already. So I
> guess everything comes down to the following question. Do you want to
> drop vibe.d or ddoc as a templating engine for the site?
> Using both doesn't seem to make any sense and for me it's not clear
> which way you want to go.

At the top level we all want a nice site, not to use or avoid specific 
technologies. We also need a site that the maintenance team can maintain 
effectively. This is something that tends to be forgotten.

We now added some use of vibe.d because (a) there was a champion who did 
the work (thanks Sönke! Also I recall Martin worked on that a bit), (b) 
it has advantages over ddoc, (c) there was hope that others will pick up 
on it.

Things have not yet reached critical mass. Giving up on vibe.d right now 
sounds like quite a bad decision. We'd lose some nice documentation, but 
more importantly we'd compromise the idea of reform of dlang.org. 
Conversely, replacing ddoc with vibe.d is not only a one-time effort, 
but effectively a maintenance burden for which we don't have enough 
qualified people.

> I remember the decision being made that vibe.d should be more tightly
> integrated into D and if that is still true the question for the
> templating engine seems to already be settled.
> If thats correct then allowing Jacob to do the work with vibe.d seems to
> bring you one step further to the goal to introduce vibe.d

Again: vibe.d is being already used at dlang.org, yet it has seen little 
pickup. The key to making vibe.d successful on dlang.org is more folks 
fluent in vibe.d willing to help on a regular basis, not more drive-by work.


Andrei