Julia User & Developer Survey 2019

We conducted the first annual Julia User & Developer Survey in June, and the results were presented by Viral Shah at JuliaCon on July 23.

Special thanks to all who participated in the survey!

1,844 Julia users and developers completed the survey from over 90 countries and fluent in more than 60 languages.

Key findings and methodology are below.

Most Popular Features / Biggest Problems

The most popular technical features of Julia are:

1. Speed, performance

2. Ease of use

3. Open source

4. Multiple dispatch

5. Solves the two language problem

The most popular non-technical features of Julia are:

1. Free

2. Julia community of developers is talented and active

3. Easy to create packages

4. MIT license

The biggest technical problems with Julia are:

1. Packages aren’t as mature or well-maintained as required

2. It takes too long to generate the first plot

The biggest non-technical problems with Julia are:

1. Colleagues, company or collaborators use other languages

2. Not enough Julia users in my field/industry

Reasons for Trying Julia

Julia users and developers say they first tried Julia because:

1. Julia seems like the language of the future

2. Faster for the work I am doing

3. I like learning new languages

4. I heard about Julia from friends or colleagues and wanted to try it out

5. Preferable syntax to other languages

Julia Use

Most users and developers use Julia for research (73%) or individual work (54%). 15% use Julia for development as part of a team and 10% use Julia in production for business critical task.

45% of Julia users and developers use Julia for at least half of their work.

Julia Packages

The most popular Julia packages are:

1. Plots

2. DataFrames

3. IJulia

4. Distributions

5. DifferentialEquations

6. PyCall

7. Flux

8. JuMP

9. Revise

10. Optim

11. ForwardDiff

12. Gadfly

13. FFTW

14. StatsPlots

15. Images

16. CUArrays

Most Julia users and developers say the package environment is ‘somewhat’ robust.

Editors / IDEs

The most popular editors or IDEs are:

1. Atom

2. VS Code

3. Juno

4. JupyterLab

5. Vi/Vim

Julia in the Cloud

The most popular cloud solutions are:

1. JuliaBox

2. AWS

3. Google

4. Azure

Julia Community: 76% say the Julia community is very (47%) or somewhat (29%) helpful and collaborative. Only 2% say the community is not very helpful and collaborative.

JuliaCon: 12% have attended JuliaCon, 59% plan to or would like to attend and 23% are unlikely to attend JuliaCon in the future.

Julia Download and Installation: Most Julia users and developers (70%) downloaded binaries from JuliaLang.org. 17% compile Julia from source.

Accelerators: Most of those who use Julia with hardware accelerators use Nvidia GPUs.

Demographics

• 60% of survey participants are academics while 43% are professionals (some are both)

  • Among academics, most are graduate or postgraduate students or researchers (56%) while 34% are instructors and 11% are undergraduates

  • Among professionals, most are engineers or developers (61%), 45% are researchers, 20% are analysts and 11% are managers

• The most popular fields include:

  • Data science, statistics

  • Engineering

  • Machine learning

  • Computer science

  • Physics

  • Mathematics

  • Artificial intelligence

  • Signal and image processing

  • Optimization

• Age: Most respondents are age 25-45, but a substantial number are age 50+

• Country of origin and language: Despite the fact that the survey was conducted and publicized only in English, respondents come from more than 90 countries, including the US (22%) and Germany (9%) and are fluent in more than 60 languages

• Ethnicity: 66% of respondents identify themselves as white, 13% as Asian, 7% Hispanic, 2% Middle Eastern, 1% black and 13% declined to answer

• Gender: 85% of respondents identify themselves as men, 3% as women and 12% declined to answer

• Sexual orientation: 4% of respondents identify themselves as LGBTQ, 79% do not and 17% declined to answer

• Underrepresented in science or computing: Among the 81% who did not decline to answer, 20% identify as underrepresented in science or computing because of one or more of the following: race, ethnicity, national origin, religion, income, socioeconomic status, education level, parents’ education level, age, gender, sexual identity or disability

Methodology

The survey was conducted June 12-26, 2019. 1,844 Julia users and developers participated, with a margin of error of +/- 2.3 percentage points. We recruited participants online using Slack, Discourse, Twitter, email, JuliaLang.org and JuliaComputing.com. The survey was administered in English, but more than half of respondents come from non-English speaking countries. Respondents come from more than 90 countries and are fluent in more than 60 different languages.