Re: Replacing the ML with a forum

From: Vinnie Falco (vinnie.falco_at_[hidden])
Date: 2023-04-26 14:43:32

On Wed, Apr 26, 2023 at 7:12 AM Marshall Clow via Boost
<boost_at_[hidden]> wrote:
> > Are you using a threaded e-mail reader? If so, do you care if
> > discussions are changed to be flat (like Discourse)?
>
> Yes, and Yes.

The current situation is this. For various reasons, we have no forum
for the new website. And a new forum is crucial for attracting new
people, as no one in their twenties or earlier wants to use a mailing
list. However, we also recognize the importance of preserving the
mailing list interface and threaded discussion model. After spending
months of doing tests and looking at existing solutions, we have
reached the conclusion that the only way to bring maximal satisfaction
to the community is for us to implement our own web-based forum which
supports threaded discussion and also has a mailing list interface,
such that the mailing list experience is unchanged for those who still
want to use it (which seems to be quite a few people).

The other alternative is to do a clean break and deploy something
completely new, which would be Discourse. The downside of Discourse is
that it does not support threaded discussions, and because of their
philosophical objection to hierarchical conversations, it most
certainly never will. Check this out:

    https://community.lsst.org/t/understanding-and-using-discourses-flat-threading/150

This rationale sounds very good on paper and it probably works for
things such as support fourms. But it is poorly suited for the
specific nature of the Boost mailing list where discussions can rage
for months (I miss that). Enormous discussions were the bread and
butter of the social experience and that needs to be revitalized.
Anyway...

Pros of Discourse:

* Already written
* Has an email interface
* Can be restyled to match the site (mostly)
* Probably cheaper long-term

Cons of Discourse:

* No threaded discussion
* Written in Ruby, a dying language

---
Pros of Writing It Ourselves
* Uses Python, a popular language
* Preserved mailing list experience
* Threaded discussions
* Customized to suit our needs
Cons of Writing it Ourselves
* More expensive long-term
* Takes longer to reach feature parity
* We have to maintain it
Note, the C++ Alliance is covering all of the expenses for development
so the expense issue is not of particular importance, our priority is
on getting things done perfectly for users even if it costs a bit
more.