Re: [boost] [review queue] Proposed new policy to enter the review queue

Subject: Re: [boost] [review queue] Proposed new policy to enter the review queue
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2017-03-17 13:12:20

> Just looking at it from a more outside point of view I would say:
> The Boost website does not look sexy.
> 
> It looks quite old-fashioned, has a lot of text, but how that is
> structured is not easy to grasp by a short glimpse. And except for
> finding the current download and the list of current libraries it is
> quite hard to find particular information fast (if at all). (The GSOC)

You'll find plenty of this already reported in depth by me in posts
past. I believe there was even a C++ Now talk or two by me on this
topic. Most of current Boost infrastructure stopped being developed
further from about 2009 onwards mainly due to a mix of lack of
volunteers to do the work, and because any attempt to make any
significant changes to infrastructure runs into major admin hurdles.

For example it took Michael Caisse nearly a year to regain control of
the boost mailing list servers and upgrade the mailing list software.
That was near a full year of tilting at that windmill, and he's a
steering committee member with all the power and authority that comes
with. It's an enormous time and effort commitment to do anything
regarding infrastructure.

That's why the Boost website and Trac install look like they do. The
Trac install 0.12.2 was last updated in *2011*. No less than FIVE
security patches have been issued against that branch since, yet to even
upgrade it to 0.12.7 which is a security patch has proven to be hard.

> Or another example is Trac. It is really ugly and leaves the impression
> that not many issues are worked on. (We had this topic already in this
> mail-discussion.)

Trac actually has really great github and git integration, but you need
to use a Trac which is *three generations* newer. Latest Trac still has
an ugly UI, but it'll auto sync up those github issues for you,
providing a unified issue reporting interface. You can report your
issues on Trac or on github, and they all sync up automatically.

It'll also port all old trac issue onto github issues for you,
preserving information, and it can match up git SHAs with wikis and a
long list of other very useful stuff.

Finally latest Trac doesn't need custom logins. You can use any of your
Google account, Facebook account, LinkedIn account, whatever.

> There is a reason, Github is so successful and people are using it (and
> in general prefer using it) despite the fact that Git is popular:
> It is easy to use _and_ it looks sexy.

I don't trust github as far as I could throw them, and I entirely expect
github to ossify and get replaced by another in time just as sourceforge
was. Trac would provide us continuity for when that happens.

> Probably, it would be a good idea to use some of the money Boost has to
> pay a professional web-designer to create a new, easier to use and
> appealing (aka "sexy") website. And then it should probably directly
> integrate the Boost Library Incubator.

I proposed that on multiple occasions over the past few years. There
wasn't consensus here for it, and the steering committee voted it down.

> And if it were possible to rate the libraries (from the Incubator) with
> a simple (star-)system and allow reviews/comments directly via the
> website (both similar to Amazon) it would probably gather more interest.

Couldn't agree more. But C++ developers still tend to think that web
development is easy and you get quality volunteered work for free. As it
used to be twenty years ago, but not in the last five years. Good web
developers are expensive, and maintaining web systems is expensive.
Boost could easily afford to pay for this sort of admin, but neither the
community nor the leadership want it and have repeatedly rejected the
idea consistently.

Niall

-- 
ned Productions Limited Consulting
http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/