What features do you want to see on Julia Observer?

It’s been a couple months since I first released juliaobserver.com.

Since then, I’ve been gradually fixing bugs and expanding features.

As of late, though, development (and traffic) have started to decline.

Therefore, I’d like to ask the community:

  • What do you want to see on Julia Observer?
  • What works? What doesn’t? What’s only so-so?

I think that declining traffic is to be expected — the ranking is interesting, but not something I would want to recheck frequently. Don’t measure the usefulness of something by traffic; you can create a valuable resource without people revisiting it on a daily basis.


First time on the site, wasn’t too clear what metric used for each tab. Perhaps these could be configurabe, described more clearly.

1 Like

What’s the functionality and purpose of Julia Observer?

And on this note: it serves its purpose. If it was any more complicated, it wouldn’t. This resource is really good, but by its nature I’m not going to check it often. It’s a great resource! But I don’t know if there’s much else to do there. I would love to see the next problem you solve!


I think Julia Observer would benefit by some documentation, including a description of its purpose and functionality, either in a brief blurb at the top of https://juliaobserver.com or under About and Help buttons.


as @John_Gibson mentioned, I’ll definitely add an about/help page.

  • but could you elaborate on how to make sorting configurable?
  • and if hot didn’t exist, would it still be unintuitive?

The main thing I want is package search, but Julia Observer seems to miss a lot of relevant packages. (Search is a hard problem.) For example, if I type interpolation into the search bar, it misses packages that show up on the first page of a general google search. (Dierckx, ApproxFun, etc.)


I would like to see the context of the package, shown in a graph, with its ancestors (things it depends on), and it’s decendants (things) that depend on it, shown.
(Probably with links).

Example shown below: (Enjoy the aweful inkscape skillz)

Potentially links could be shown between other nodes in the graph (which I think would be interesting).
(This would be the induced_graph between the set of ancestors unioned with the set of decedents)

I think if proper graph layour was used this would look a lot better.

The point of this view is to have a good idea of what would break if a package broke, and what breaking would break this package etc, and to generally see where the package lays in the ecosystem


Sweet! Add perhaps a version since these are commonly restricted in REQUIRE.

1 Like

A warning: maxima-ecl, (manjaro, arch linux) does not work.

Nice changes! Like the Categories.

I see from the About page you’re drawing inspiration from other package browsing tools. One that I use often and think is well designed is https://www.rdocumentation.org/ by DataCamp. Perhaps there are some features that that you might find interesting.

1 Like

Where does maxima-ecl live?

For julia observer to catch it, there would have to be a github repo named:

  • maxima-ecl.jl

Just a heads up, all proposed changes are now live.

Let me know what you think and possibly throw out some new ideas

1 Like

As a note, when swapping between the categories and news tab it takes about 3 seconds per swap for me. Makes it feel pretty sluggish.

1 Like

This stems from the fact that the app’s built on rails (and not using a javascript framework).

// i.e. switching between categories & news initiates a server route change

A few questions:

  • Was the initial load slow?
  • Were subsequent loads slow?

Also, was it slower going:

  • from categories to news
  • or from news to categories?

// edit: one solution would be to do some caching on the actual pages (would speed up search too)

Initial load is quite fast. Going from categories → news or news → categories takes the same amount of time.

I wrote a Plots recipe for this at some point

There is an example of it in use here: Scope in metaprogramming - #17 by mkborregaard

1 Like