[peatmoss]

Over the years I’ve kept coming back to Julia, like what I see, but ultimately end up with a deadline that has me reaching for R or Python. My latest foray has convinced me that the ecosystem is ready for me to double down and make it my language of first recourse for data science related tasks this year.

Precompilation sounds like it improves further on an annoyance (time-to-first-plot problem) that was already no longer much of an annoyance in Julia 1.8x.

[DNF2]

Just to manage expectations: As far as I understand, v1.9 doesn't by itself solve TTFP/precompilation. The tweet seems slightly over-enthusiastic. But it puts tools to solve it in the hands of package developers.

In that sense, one can perhaps say it is a turning point.

[ViralBShah]

While that is an accurate statement, my go-to package where compile time and startup time was a huge hurdle, Plots.jl, just works instantaneously now. That is the case for many packages already, and I suspect that after 1.9 releases, it will quickly become more commonplace.

[peatmoss]

Ah, thank you for that context. Assuming this mechanism doesn’t have any major tradeoffs, I think it seems likely that popular packages will make use of it?

[ViralBShah]

Two tradeoffs to be aware of:

1. Slightly longer compile times at package install time, but that is ok since it is a one time cost.

2. During package development, building a pkgimage every time will lead to longer compile times because of native code generation. That is easily disabled during development with `julia --pkgimages=no`.

The popular packages are already preparing for this.

[kelipso]

It's very likely. I've only occasionally used in julia for a couple of years and I've been doing various hacks to improve load/compile times for my private packages pretty much since the beginning, so I imagine more experienced programmers would jump at it.

[cbkeller]

Yes, I think so (and many already have)