Re: A few questions: Libre SoC, website, Rust

[Samuel Thibault]

Jan Wielkiewicz, le dim. 16 août 2020 21:00:26 +0200, a ecrit:
> > More than a grant, it's people that we would need (yes, a grant could
> > help here but in the end it's people that matter).
> I believe the website could help here, explanation below.

Sure. But also see below.

> > Being able to work on people's hardware is also a very important
> > thing. You won't attract people to your OS if it can only run on some
> > "obscure" hardware. Supporting x86 remains some must, and porting to
> > 64bit will be the most efficient way of fixing the pending year-2038
> > issue.
> I see. I'm not sure if the Hurd's ideal target are x86 devices anyway.

Whatever the "ideal" target would be (we have previously seen people
saying arm would be), not targetting x86 as well looks like suicide to
me.

> Are companies going to write proprietary firmware for the Hurd in the
> near future? Likely not.

Why would they need to be? A firmware is not supposed to be OS-specific.

> People who want to run only free software on their hardware are
> already forced to buy some obscure hardware

Ok, but I believe we do not want to target only them.

> By providing Libre SoC support for the Hurd, the project can prepare
> for *new* hardware that's already comming, instead of trying to chase
> 10 years old proprietary junk.

We are not really chasing for hardware support, we just use existing
drivers (Linux, OpenBSD, etc.)

> > It depends what you call "modernization". Putting javascript etc. is
> > basically a no-go. Redesign the css etc. would probably be useful
> > indeed.
> We're not going to add some unneeded bloat, just some CSS eyecandy and
> actual features. Also answering your later message, I'm aware of "the
> JavaScript Trap" and this is not a big issue here - all we have to do is
> to release our scripts (if any) as free software and to add some
> metadata on to the website so GNU LibreJS knows the scripts are free
> software.

Then that'd be all good.

> > But also we need somebody to keep up with the
> > quarter-of-the-hurd just to mention on the website what is actually
> > happening.
> Currently the latest news on the website are "2016-12-18-releases".

Yes, the main website is lagging behind. I don't really know
how it is supposed to be updated, Thomas knows. Please bug
hurd-maintainers@gnu.org about it. I have now pushed at least the news
part.

> > I will probably never understand the reasoning behind "graphically
> > appealing => is alive", but I guess that's the world we live in.
> How we present the project to the world and its appearance is extremely
> important to success of the project,

I know. That still does not mean I understand the *reasoning*.

> a good example of this can be seen in the animal world:

We are not animals.

> Some animals pretend to be dead in order to trick predators into
> thinking something is wrong with them:
> https://en.wikipedia.org/wiki/Apparent_death

Non-graphically appealing doesn't mean "dead".

There is no reason why we should have to chase the latest-shiny-brighty
design trends (which basically means revamping the website every year or
two) only to express that the project continues.

I know, the monkey prefers the red car. But we are not monkeys.

> People are simple - they see an old website,

I'm afraid I'm not sure I want to attract people who only like
shiny-brighty websites and can't stand a merely black-blue-on-white
design. By this, I mean: personally I just ~#{[ don't have the #{[[ time
to answer their probably terribly large amount of questions.

> with latest news from 2016 and assume the project is dead.

That part makes full sense, yes. Now fixed. Again, as I mentioned, it's
a matter of somebody actually taking the time of putting news there. I
don't have time to do such a thing, and I'm no good at that anyway. I
can proofread what somebody would write, though.

> professional website = people thinking the project is professional

"professional" is a very relative thing. For me black-blue-on-white
looks more professional than shiny-brighty. But again I'm probably
simply just biaised and just not adapted to the world as it is nowadays.

> By just saying on the main page "we're looking for developers for X" we
> can gain new contributors.

It is actually written there.

> This is the power of presentation. Really, I had to tell people all
> around the Internet, the Hurd is still alive, because they thought
> otherwise!

The Hurd carries a huge background of urban legends etc. and most people
don't bother actually checking for themselves, so I'm not surprised at
all (and no, I don't think things were different in 2016 when the latest
news was fresh).

> > It is already a wiki.
> Yes I know, what I mean by this is we could make the main page with sole
> purpose of showing the Hurd is cool and still alive and a candy shop
> people want to come and from this cool main page, there would be a link
> to the wiki, the current web page we don't want to show anyone but
> maintainers. This would save the current maintainers from headaches.

Which headaches precisely?

Making the page cool could attract people, yes. These will however have
a lot of questions (they won't bother reading the FAQ). Personally I
simply won't have time to answer. If people here on bug-hurd and #hurd
do have, then sure go ahead.

I'm sorry I'm being very negative here, I know that's definitely not the
way to run a cool and fun project. The thing is: I never asked for being
in charge of doing this, and the other thing is: the most badly needed
things that we lack is not really that much fun and cool.

Just for instance, the rust support:

> > Yes, some things could be done in rust. But rust would need to be
> > ported first. 
> By this you mean it should just run as on GNU/Linux or are there some
> *special Hurd steps* to port a language to it?

Just as on GNU/Linux.

> I can check if Rust runs already on the Guix Hurd images.

It just doesn't. 

> > I did have a look, and the work needed there
> > (basically, explain rust the API of each and every function provided
> > by libc, while rust could merely read it from the .h files)
> > discouraged me quite a bit.
> I see, I'll ask my friend if he understands this, he's a better
> programmer than me.

Just to give you an idea, read
./liblibc/src/fuchsia/mod.rs
which is the Fuchsia file needed to make rust work on Fuchsia (there are
others, but that's the main one).

That's 3950 lines of code, which are neither fun or cool, but that we
apparently *have* to write to make rust work at all on the Hurd, because
it seems that rust maintainer never bothered to write something that
would simply somehow generate these lines from the libc headers like
compilers for other languages do.

I don't think making the website shiny-and-brighty will attract people
who will be patient enough to write such a file.

Again, that's only my own opinion, and what my limited amount of free
time can permit, I'm sorry that it looks so negative. If other people
can help with this, feel free, it's an open project, I'm just warning
what I have seen happening in the past decade.

Samuel