[ocaml-infra] Redesigning OCaml.org

[Amir Chaudhry]

Hi folks,

Thanks for the useful discussion!  I've tried to collate my responses here rather than replying to the individual emails.  I've tried to cover all the points but please let me know if I missed anything.


## OCaml logo

It's taken a while but I've heard enough comments (from a variety of people) that they really don't like the lettering on the new OCaml logo [1].  I don't recall anyone stepping forward to say they love it.  

As such, I'll take this aspect of it back to the design agency to reconsider and I'll start a *separate thread* to ensure I've got up-to-date feedback.  

Thank you for voicing your opinions. I'm very grateful.  It's difficult to please everyone but for the logo, we should try and avoid things that people strongly dislike.  I'm expecting it to be around for quite a while so it's important that people in the community are willing to use it. :)

[1] https://github.com/ocaml/ocaml.org/wiki/Draft-OCaml-Logos


## Pages on the demo site

Please note that there *are* actual pages to look at e.g the 100 lines of OCaml [2] and tutorials page [3].  Just click on some of the links at [4].  These are not the final versions (in terms of styling) but I'm a bit concerned that people might only be commenting on the front page and nothing else.

[2] http://philippewang.info/CL/new-ocaml.org-demo/taste.html
[3] http://philippewang.info/CL/new-ocaml.org-demo/tutorials
[4] http://philippewang.info/CL/new-ocaml.org-demo/menu.html

### On use of Lorem ipsum

There's only lipsum on the new pages that will help visitors find their way to the deeper content (e.g landing pages for the main menu items).  I made an effort to come up with initial headings so the only lipsum is body text.  The problem with lipsum is that the visual look of the page is altered when you replace it with actual text.  Things like word length and line-breaks affect the overall 'balance' of the content, which can sometimes be troublesome.  Having gone through that once before on other projects, I'm already aware of the issue.

Despite that, I disagree that we need proper text to assess the design.  If it doesn't work with lipsum, then adding text will *not* miraculously save it.  Using lipsum this way also gives you a feel of whether there is too much text, i.e. I now think the platform page has too much [5] and that's only around 70 words per paragraph.

[5] http://philippewang.info/CL/new-ocaml.org-demo/platform.html


### On drop-down lists

I'm really really not keen on drop down lists.  I can understand why they're useful but in my mind this is a website, not a user application.  As such, I do not think they belong in the primary navigation bar.  

Using drop-downs also makes an implicit assumption about the visitors, which is that they already know what they're looking for and where it lives. I don't agree with this assumption (of course, everyone here *does* already know the site, but there are many more people out there who do not).

You can look back at the Scala site [6], mentioned earlier in this thread, and see that they also avoid drop-down navigation lists.  So do a lot of web-apps that I'm familiar with.  

Please also bear in mind that most web-users find their way to content via search engines.  We shouldn't assume that all content is discovered by travelling from the first page onwards, hence I don't consider 'click-depth' to be a relevant metric.  That's fine for a collection of categorised bookmarks but we're aiming higher for the public face of our language.

An alternative, already mentioned, is to have more info in the footer.  I think this a good suggestion which could be done sensibly and avoid crowding the main navigation, though I'd think carefully about which links to put there (it shouldn't be everything).  Suggestions welcome :)

[6] http://www.scala-lang.org


### On the use of white-space and density

This is related to comments above.  *Reducing* the density of content and effective use of white-space makes things easier to navigate (there's less to mentally process before making a decision). 

This is also why larger fonts are better for the website.  When you know your way around something you appreciate density and the ability to see more at once e.g. think about the font sizes you use in your terminal windows or favourite editors - environments you've likely customised and are *highly* familiar with.  The same approach isn't appropriate for a website, where the audience covers a wide range and aims are different.

Having said the above, I do agree that the icons are a little large but they break up the page well.  I haven't made up my mind about the 'learn' and 'platform' icons but I do think we should have them. The navigation bar is also a bit too large and I'd like the news section to be a more prominent.  I'm wondering whether changing the first row to a four-column layout may be better.


## Package information

I completely agree that 'popular' is a misnomer.  'Most downloaded' makes way more sense. I did find the list useful when I saw it in OPAM as I had no idea that so many things depend on ocamlfind.  Of course, now that I have that knowledge, that particular tab wouldn't be much use to me (as the content wouldn't change much), but that doesn't mean it should be dropped altogether.

I do like the idea of a more detailed page elsewhere that shows more stats from opam e.g dependencies, (meta)data from OCamlot, etc. but they're not necessarily front-page items.


## Aside: Good designs last forever?

Designs change.  I'd argue that logos last a long time but that's because there's usually lots of investment and brand recognition attached to them.  Think about the Coca-cola and McDonald's logos.  

That good *brands* can stand the test of time should not be confused with good *design* somehow being unchanging.  

Best wishes,
Amir