I'm like you Seth in that I don't use most of the stuff that people on this 
list talk about. I also have no frame-of-reference to MATLAB, just a 
Python/R/SQL guy (and recovering SAS user).

For me, the language is plenty stable, but I also appreciate a challenge. 
But for my common workflow, I mostly just need to interpolate strings, get 
data from databases, make plots. I successfully use Hadoop via Julia via 
ODBC via Hive (whew!) nearly everyday, and it works great for me. Gadlfy is 
a bit slow to get started up, but given that I can output to d3 and post to 
the web...that's a pretty killer feature. 

So it's really just perspective. I show people what I'm doing, and if they 
get interested, great. As a previous commenter wrote, there are people who 
create languages/packages and people who wait for the packages to be 
developed. 


On Thursday, March 5, 2015 at 10:38:00 AM UTC-5, Seth wrote:
>
> I'm probably one of the few Julia users who's NOT using it for scientific 
> / numerical analysis / very-obscure-technical-field work. I'm just a 
> general programmer - I'm more interested in getting data from point A to 
> point B, processing it quickly (whether that's parsing web server logs or 
> creating user activity graphs - a lot of the work I do is security 
> related), and then outputting it in a way that makes sense to me. I use 
> python but find that Julia's speed (both of execution and of development) 
> is too good to pass up, especially for complex things. (Also, python's 
> regex handling is horrific, and really, no general programmer is still 
> using Perl :) )
>
> For my use case Julia is both amazingly good and incredibly frustrating. I 
> love writing Julia code, but the instability of some of the core libraries 
> drives me nuts. Here's my biggest challenge:
>
> Core to any language should be a standards-based way of getting data in 
> and out. Right now Julia's a bit of a mess in that regard. For example, 
> Base has download(), which in turn relies on system utilities: curl, wget, 
> or fetch, according to the docs, which also say something to the effect of 
> "casual use only". JuliaWeb (of which I am a member) has a variety of 
> packages with duplicate functionality, including Requests.jl and 
> HTTPClient, but there is no settled standard yet for the underlying 
> technology, and each package has its own set of limitations not found (or 
> better hidden) in more mature packages in other languages.
>
> Until we figure out a way to make it easy for folks to get their data into 
> and out of Julia, I think we're leaving potential new users "on the table", 
> unwilling to switch because migration is just too hard, and the lack of 
> good data transfer / network utilities is a large part of that. So this is 
> my plea: if you're interested in helping form the basis for a great 
> web/data transport stack in Julia, please join us in JuliaWeb. We've got a 
> list of issues posted https://github.com/JuliaWeb/Roadmap/issues in case 
> you feel an urge to help fix some things and develop some new things. It's 
> pretty much a green field at this point.
>
>
>

Reply via email to