As we’re getting to the two year mark of JuliaObserver, I thought it was a good time to check in w/ you guys.
Currently, the website has ~1k weekly users, with an average session time of 3-4 minutes.
If you could, please fill out the following poll. All comments are welcome, so be as honest as you want.
0 voters
I like Julia Observer, but i kind of miss pkg.julialang.org/pulse.html
I liked the first graph reporting number of packages per version and the 14-Days stars change. These may be quite easy to add to julai observer though.
I also feel like julia observers design is more suitable for big screens or maybe touchscreens. I’m on a 14 inch screen and would like a higher information density.
I like the Observer and it’s definitively a very useful service. It also got an extra responsibility when got linked directly from Julia home page. So the information it displays should be the most accurate as possible but it turns out that it’s difficult to make corrections. I’ve opened an issue some months ago to fix a bit of that info (a package classification issue) but nothing happened. I understand that it relies on julia.jl to get the info (where I also submitted a PR) but the fact is, no update took place.
In summary, my point is that it should be easier to improve the quality of the information displayed in the Observer
I like the observer. I wish either the observer or julialang/pkg would attempt some editorial guidance. See https://discourse.julialang.org/t/pkg-ecosystem-learning-from-others-mistakes/18028/51 for a long discussion.
What it boils down to is some human-maintained subset of “high quality” packages, even if this ends up as a super subjective and unprincipled directors cut.
I am unsure about its usefulness. I am missing a very simple but large(huge) page, with every package available/registered just to be able to use the browsers text search function. This fancy javascript auto line adding when scrolling down does not give any better usability (it’s by the way annoying in discorse too). And I guess it is not the best for being indexed by search engines. So this is an example for responsiveness without benefit. There are still double entries in the categories (e.g. Statistics, i didn’t check for others).
So all in all I think it should be less mobile/responsive/fancy and more simple with “higher information density”, to quote above.
What I like about it especially is the reverse dependency list, that is, for the current package, show where it is used. This information is not so obvious from browsing raw GitHub.
When searching for packages a scroll down box is presented with packages meeting the search requirements
Other than these two nits JO meets my needs.
I love it! I like learning a package I’m new to by finding other packages that depend on it and looking at their implementation. Observer is great for this. It does get a bit slow in my browser but I’ve also been known to get frustrated with how slow my microwave is
JuliaObserver is very useful, it’s essential to have a view of packages by category and sorted by popularity (else you get lost given how many packages there are, including unmaintained ones). Though categories are not always right, and it’s a bit inconvenient to rely on Julia.jl for that. I was thinking you could extract the GitHub tags too? That way maintainers can directly control how their package is classified.
Did observer suddenly get faster? I just noticed it was working really quickly this morning.
There probably were just fewer users on because it was broken the past couple days.
// upgraded the servers a few months back, but it shouldn’t be that noticeable 
The categories are not good. For example the categorie “Control” is missing. There are lots of good control packages for Julia, see for example JuliaControl · GitHub .
There could also be a categorie “Control and Simulation” which then should include Modia.jl and many others.
And it’s also very difficult to interact with the system. I’m trying to reclassify GMT.jl since last September but not successful yet.
@joa-quim, I just pull in information from Julia.jl’s db.csv file. I would advise making a PR to update that file?
The whole purpose of the website was to collect information (on Julia) from N sources, ~5 times a day. I tried to distance my own biases and moderation from the process.
What is this db.csv? How is it created and maintained? And by whom?
You mean this one?
Hi @djsegal ,
Heres to wishing you luck with Juliaobserver.com 504 error ( gateway timeout ) .
And here’s a first shot at the issue per AWS AWS offers solution to resolve 504 Errors with Cloudfront CDN
If the site has been hacked down what is Amazons AWS SLA to restore service ?
Good Luck and message me if I can help you.
I miss JuliaObserver
Hi @oxinabox ,
Its been down over a week now, and its not clear if anyone is working on it .
All the alternative “work arounds” are extremely cumbersome and weak substitutes;
Would you or anyone else on this thread know who to contact, offer help to etc. ?
I think it may have intentionally been allowed to lapse,
on the theory that the new and enhanced pkg.julialang.org could take its place
but it is yet to get to be as good.
As I understand it JuliaObserver was effectively frozen since the registry change over in April.