How do you push to production?

Specific Domains Data
1

So, when I write some algorithm for processing data in R, after testing, the act of pushing it to production is usually little more than basically having a pre-existing application or process call the R script.

But, since R is an interpreted language, there isnt much more overhead in efficiency between my testing environment and when it’s put into production.

For Julia, though, because the first run every time it’s called requires compilation, just having the application call the script I wrote will cause a significant amount of overhead, because each individual call requires compiling again.

So, what’s the best way you’ve found to push your Julia scripts/function to production that doesn’t require compiling it at every call?

1 Like
2

This is essentially the “first time to plot” question. For example, running plot(rand(10)) takes a few seconds for the first run as it needs to compile but its blazing fast after. As a package author, there isn’t much you can do about this. The devs have made it a priority to address the long compilation times and version 1.5 has significant improvments.

1 Like
3

See Modules · The Julia Language

1 Like