the website is completely maintained by volunteers, as such you can open issues on GitHub with specific ideas (ideally PRs) of things you think could be improved
would you rather $$ was spent on making the website a bit sleeker or on, say, improving the compiler, growing community etc?
(2) is a semi-trick question, you could argue that a sleeker website could attract more people but it’s not straightforward to prove, IMO people come to Julia because they hear it might be a good language for their use case, not because the website looks cool, as long as it’s functional.
While I don’t fully agree with @Thibaut, I feel him. IMO, our website certainly doesn’t look as “stylish” as it could. Among the new talked-about programming langues (Go, Rust, Julia) I probably like https://www.rust-lang.org the most (although I dislike the font size and a few other things). I guess what I am missing on our website is (background) colors.
But even more than the design itself, I think we could improve on the presentation of our content (which is why I love to hear that the download page gets a revamp). Personally, I feel there is in general too much text. In particular, text blocks that seem redundant or unnecessary that just make the website look overly complex.
For example, I don’t like the “start” of the downloads page:
The first thing we do is ask for likes, which, to me, seems a bit rude. Then we point out that there are docker images and Julia Pro. IMO, this is extra information and nothing that should come before the actual download of the Julia binaries. Next, we talk about potential installation troubles. Why? I haven’t downloaded the binaries yet. Then comes a privacy note about the package manager. Again, personally, I find the positioning of this note way to prominent. Not because it’s not important (Frankly, I personally couldn’t care less ) but because I still haven’t seen or downloaded the binaries without which I can’t even access the package manager.
I really like the idea of replacing the binaries-table and the following (overly complex) text block with a simpler download button and more subtle links to nightlies and old releases.
While this is true, I think that a great overall design (not just the images and colors but also of text and information) is hard to establish through iterative contributions from many individuals. The latter can refine things but needs a solid fundament. Of course, one could do and contribute a redesign on its own.
Yeah, I have been a little provocative but this is for the sake of good. I really like Julia and I think that an unicorn needs to have a site that it deserves.
The idea is not to make something beautiful just for the beauty. The idea is to transmit some messages, like the fact that Julia is now a mature language.
The impact of the visuals on the brand identity is not to be proven. It has an impact in the brand perception that goes trough conscious and unconscious processes.
In case of doubt, it also works on engineers even if many will never admit it…
I worked on an open source project with the french Design Research Workshop (ARD) ten years ago. I can tell that it was very hard to find designers willing to use them free time on an open source project due to the cultural gap and the fact that most of them are freelancers who already have a hard time hunting for paid projects. I hope it changed a bit. Otherwise this may be doable to have a first graphic canvas for something like $5K.
While this is not ideal hybrid people like me who are able to make correct things could make proposals. I can do one or two later on on the beginning of the year when I will have finished some urgent work. Of course this will never be as good as what could do a talented graphic designer in a few days…
I agree that there is also a work to be done on the site readability. The idea would be to think about the main messages and to make it impactful. Even working with a skilled UX designer this would mean to have some people with a high level of knowledge of the site to make some preparation work and to think about the messages. I’m pretty sure that there is marketing specialists among the data scientists of the community who know perfectly what I’m speaking about.
It is usual that the need for such communication oriented design raise some doubts in a technology oriented community. However this is not something that is trendemously difficult to do once you have the right people. The goal is not to make something incredibly beautiful (even if this would be cool) but to transmit the right messages to people passing by.
I think most of us will agree that it would be nice to have a uniform design for the website; this might happen in the future if someone™ ends up spending a couple of days making a bootstrap, bulma, or whatever skeleton (I might even do that if I find the time).
One thing that both you and @carstenbauer point out though (with which I agree) is that the text organisation, length etc of some parts of the website could be improved. That is pretty easy to open PRs for; and I and others will be happy to help you out with formatting or whatever :). Anyway this is probably a separate discussion, bottom line is: open issues on the repo, make suggestions, we’ll help you do the rest
FWIW, I’ve been ~slowly~ migrating pages from full on text blocks to be more like the home page design. The research and JSoC pages (along with download) are the last ones that need to be updated.
Perhaps Mozilla and Google have (or had) a few designers on their payroll… Not sure the Julia Community even has a payroll
There’s a lot of subjectivity in design, which makes this kind of project quite challenging. It’s difficult to please everyone. There are hard-core CLI-minimalists who’d prefer a text-only give-us-the-facts approach with underlined links, zero JavaScript, and no marketing buzzwords; and then there are people who are seduced into trying new things based on big fonts, animated graphics, sensuous color schemes, and cute animal mascots. Many will have their own preferred place somewhere between the two extremes. You might enjoy reading this Rust language forum discussion about the re-design of the Rust langauge web site. Good reading before you embark on your own re-design…
For julialang.org, the first thing I’d suggest would be to implement the “Respect Dark Mode” issue. I’ve tried to implement something similar on another Franklin site, and to be honest I struggled a bit with the CSS. I’ll chip in a UK pound or two if that’ll help?
What I find a bit distracting in the current version is the splitting of rows at the word portable. This makes scanning the table a bit more difficult. Maybe a new line for a new version would be nice.
Another point is that with the new design I need to open another site which increases the barrier.
Maybe I missed this, but I think the new version combined with an automatic OS detection would be nice. The other option would be to just show the 3 most common download options and a link to the full list
Web console:
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:84.0) Gecko/20100101 Firefox/84.0
Your current platform has been detected as: downloads:1:1764
Linux x86_64 downloads:1:1825
Your current platform has been detected as: downloads:1:1864
Linux x86_64 downloads:1:1925
Uncaught ReferenceError: downloadlink is not defined
getOS https://julialang.netlify.app/previews/pr1110/downloads/:1
<anonymous> https://julialang.netlify.app/previews/pr1110/downloads/:1
EventListener.handleEvent* https://julialang.netlify.app/previews/pr1110/downloads/:1
downloads:1:1952
Source map error: Error: request failed with status 404
Resource URL: https://julialang.netlify.app/previews/PR1110/libs/bootstrap/bootstrap.min.css
Source Map URL: bootstrap.min.css.map
Personally I’m a fan of the automatic OS detection.
I actually like both the current download page and the suggested new one: both of them provide the essential information without too much else cluttering things up.
Feedback (focusing on the negative points, as is often the case):
The text below the main green download button adds visual clutter, as it is too small to be easily readable, and too close to the main button.
I assume the square brackets around “[Help]” were deliberate, but they look like a mistake. “Help” is also not a very descriptive title: it’s not clear what it links to or what kind of issues it might provide solutions for. I’d suggest “Install Instructions” instead.
I’d change “Other Platforms” to something more generic like “All Downloads”.
I think the “Why Julia” box is unnecessary. If I’m on a download page I probably don’t need to be convinced about that. The space would be better dedicated to some other purpose, e.g. a description or link to the next steps on getting started.
On Linux, you don’t want a tarball that you download then figure out how to tar -xvf ...: most people don’t know this (sure, in the biased sample of package developers people will know this, but most people in an ML workshop won’t!). So instead of just a button, I’d put step-by-step instructions or even better, just a “copy-paste and it’ll work”.
Compare the button we have to what PyTorch gives me:
I don’t even know what torchvision is, but copy-paste and it works.
So on Windows and Mac I’d like to just see a download button with the platform specific installation instructions right below it. On Linux, I would like to see a script that copy-paste installs by throwing it into the terminal. Then you should only have to hit the help button if you want to find out information about other platforms or versions.
) but because I still haven’t seen or downloaded the binaries without which I can’t even access the package manager.
