0. Warnings/disclaimers: as usual.
Things that *really* bug me with the new page:
1. The boxes on top. Pretty much what Alexis originally said
(https://github.com/racket/racket-lang-org/pull/28#issuecomment-267796721)
2. More concretely, in not-so-large screens (laptop/phone), I don't know
what I'm looking at. On a typical ADD-style web glance, it looks
like yet another "language framework" -- the typical yawn inducing JS
things, not what Racket really is. This is from "make languages".
But it really doesn't look like a language: from "solve problems"
(way too generic; applies just as well to smart washing machines),
through the rest of the titles that are visible on the top. Perhaps
the only real clue that it's a language is "The best of Scheme and
Lisp" -- and IMO/E that's not a great motivation...
3. Even on a large screen, the pictures seem to go out of their way
trying to make me think that this is something related to graphics.
4. The whole business with the boxes becomes *MUCH* worse when viewed in
a text browser. You get all the boxes text, and much later the blurb
that says what it actually is. It pretty much looks like a kind of a
quasi-random word cloud, all the way to the latter part that looks
more organized.
5. "Who cares about text browsers?" -- Consider that they can reveal
other problems, for example, the (oddly placed) main blurb text is
shown as a list item with one of them.
More severely, one of the very good points about trying a text
browser is that you get to see the page as a search engine sees it.
I initially thought that this must be bogus advice these days...
Turns out it's not: when I search for "Racket" here's what google
tells me about it:
> Racket is a mature LGPL project that's actively developed and
> maintained. Racket repositories. Main repository. The PLT Group.
IMO, this is a *MAJOR* failure of the new layout.
5a. Google also makes a mess in the quick links that it infers for
the site.
5b. For fun, I tried the other search engine. It's similarly
confused -- here's what it says:
> Racket’s crown jewel is its macro system, which lets you
> freely extend the language. Intro to macros; Macros in depth;
> Racket syntax model; Making new languages
5c. Yet another reason for using a text browser is accessibility. I
tried the chromevox thing, and it just reads out what's on the
page, top to bottom. It doesn't read the text hidden in the top
boxes so it's not as bad as a text browser, but it's stil far,
far from good.
6. Gray text. [IMO the universal default of black text on white
background was a colossal mistake propagated by the misguided idea of
making computer screens look like printed text -- and for years I
thought that this was so bad that there's no way it could be worse.
Then came the style of gray-ish texts and proved me wrong.]
AFAICT, the only black text in the whole thing is the first "Racket"
word. Section titles on the bottom part look like they should be
black too but the font is thin enough that there are very few pixels
that are actual black.
Yes, I've seen some justification about gray being a common
de-emphasizing tool, but in that case I look at the text that I would
read if I ignore de-emphasized text, and that says "Racket, Software,
Documentation & tutorials, ...".
ObYoutube: https://www.youtube.com/watch?v=Excu14T7IxQ
7. "the world’s first ecosystem for developing and deploying new
languages" -- This reads to me as a huge exaggeration. Wouldn't
Lex/Yacc fit this description? Actually, an OS (Unix, specifically)
could make this claim too. On the flip side, "enterprise people"
probably read this as something that better fits what Visual Studio,
or some JS thing like Esprima is doing.
To put this differently -- IMO the thing that makes Racket unique is
exactly the *opposite*: I don't have to "develop" a language, I can
just dig around through my back back for a paper clip and some used
gum and slap up a language, and then I most certainly *don't* need to
"deploy" it! -- IMO, that's kind of the main sentiment whenever
Matthias talks about making languages so easy that you can make a
language for a single use -- again, the opposite of common thinking
where a "language" is something so heavy that it doesn't make sense
to not deploy it in some way.
8. Finally, phone rendering makes the above problems even worse. But in
addition, it is still broken (for me, on a Nexus 6p using chrome):
http://tmp.barzilay.org/x1.png
http://tmp.barzilay.org/x2.png
Both have overlapping texts, and the first one has a strange newline
before "dream language".
--
((x=>x(x))(x=>x(x))) Eli Barzilay:
http://barzilay.org/