Databases are one of my other large areas of interest (besides strings, decimal arithmetic, performance, compilers, interpreters, among many other things!), I’d love to collaborate with you on a uniform database API.
I think it’s important that it be flexible enough to deal efficiently if possible with at least row-oriented and column-oriented databases, and hopefully also things like “NoSQL” storage, i.e. document, key-value, hierarchical, etc.
I think Julia is probably the best language currently available to make an easy to use DB API.
Big second on the importance of database support. I know of at least one hedge fund who looked into Julia, loved most of what they saw, but ultimately didn’t run with it because of the difficulty interacting with SQL databases.
Such polls are always biased to some degree. However, for pure data science tasks, Julia may not be the best choice. I really want to see a Matlab style IDE for Julia, like RStudio and Spyder, which is well suited for data science.
I think it’s kind of better too since MATLAB/RStudio are not really well-featured as text editors. It’s hard to get column edit, Vim style, etc. working in those, while Atom and VSCode are text editors first so they have all of the fancy text editing features + the IDE.
I never liked Atom due to its performance issue, and I use the VS code Julia extension. I feel like Matlab style IDE is more comfortable to use, for example when I have more than 2 monitors, I can drag and enlarge some data frames or plots for better view, and this makes it easy to teach. Now VS code Julia extension does not allow full screen plot or data frame in a new window.
A straight forward IDE is very important to attract new users, R has RStudio and Python has Anaconda distribution, and they are very easy to get started.
Another thing I do not understand is that so many people are telling me that Julia solves the two-language issue, then why not have a Julia specific IDE? I thought the reason of using editors like Atom or VS code is to make it easy to work on different program languages at the same time.
Ehm. For myself and a long list of colleagues working with Matlab’s internal editor that’s very, very low on the needed features.
What is premium and don’t see this around is the on-line linter that’s interpreting your code while you write it (e.g. ‘the value assigned to variable x might be unused’ - which can point you to typos of names) and the warnings/error bar on the left.
And there is the simple GUI for breakpoints and mouse-over for values.
I think it depends on the users. I’ve mostly not used IDE’s, certainly I’ve never relied on them. Among my data science colleagues some use Spyder and some use vim like me (I haven’t really looked into Spyder that carefully to figure out why people love it so much).
Sure, but that in-and-of-itself isn’t a reason not to use it for any specific language. vim and emacs have been around for more than 30 years, they work as well on Julia as they did on FORTRAN. I love Julia and at this point I rarely use other languages, but I don’t see that as any reason to jump from vim to anything “Julia specific”.
I’ve been using Emacs (and Emacs-like editors) since 1980, still the most productive tool for me, works quite well with Julia.
It’s basically whatever tool works best for you.
Because there is a universe of distance between them … called debugger. One has it and the others don’t. Hopefully that will change but mean while, IMO, there is no comparison.
Well, to have a good debugger just needs to update/improve the existing Julia debugger core (a lot of work, though). When this is ready it will take only weeks to have it integrated into Atom/ VSCode. So if you miss a good debugger, this is no reason at all for a Julia specific IDE.
Spyder does not have a for profit organization support as Julia does, so I guess rstudio should be more comparable. I don’t think spyder will be dead. Even gnu octave survived, spyder has more users, so I think it will get over this ultimately.
I’m here to agree wtih @Yifan_Liu. I came to julia from matlab, which was the first language I learned, so switching away from an IDE was a jump for me, and I still use matlab for a lot of things, especially instrument interface.
However, I agree that a “real” text editor is important. I don’t even do the majority of my matlab editing in the matlab editor! Instead, I write a big chunk in sublime and then open it in matlab to check for typos with the linter.
Fortunately, not having a linter in julia hasn’t been too big of a deal for me, since most of my typos are using julia-isms in matlab code.
