How to upgrade Julia to a new release?

The only manual step I have noticed is, if you work with IJulia, you I think I needed to go ] build IJulia prior to working in jupyter after the installation.

2 Likes

sudo rm -r /usr/lib/julia-1.0.1
sudo mv /Downloads/julia-1.0.2/ /usr/lib/
sudo ln -s /usr/lib/julia-1.0.2/bin/julia /usr/bin/julia
julia

6 Likes

On Mac OS, you just install the latest dmg file and replace the older one when it prompts you. Of course, this won’t work if you are moving from v1 to v2 as the path is major revision specific i.e. /Applications/Julia-1.0.app/Contents/Resources/julia/bin/julia

1 Like

This is on Windows?
I recently installed Julia 1.0.2 on a system which originally had 1.0.1 installed.
Just install the new version.
The only thing is that a new desktop icon was not created.
As above, there is semantic versioning 1.0.1 to 1.0.2 is a patch release.
The 1.0 is the interesting version

So just install 1.0.2 GO into the package manager and type status
You will find you have your original set of packages.

Please stick with Julia. I have gone through a lot of the feelings you have, regarding corporate proxies and versioning.
Please dont get frustrated. We can help!

4 Likes

installed julia-1.1.0. Run the pkg> st command

(v1.1) pkg> st
    Status `C:\Users\hermesr\.julia\environments\v1.1\Project.toml`
  [c52e3926] Atom v0.7.14
  [e5e0dc1b] Juno v0.5.4

(v1.1) pkg>

Only two packages appear.!!!

In julia-1.0.3 are these packages:

(v1.0) pkg> st
    Status `C:\Users\hermesr\.julia\environments\v1.0\Project.toml`
  [28f2ccd6] ApproxFun v0.10.3
  [d9fa014c] ApproxFunExamples v0.1.0 #master (https://github.com/JuliaApproxima
tion/ApproxFunExamples.git)
  [c52e3926] Atom v0.7.12
  [ffab5731] BlockBandedMatrices v0.3.2+ #master (https://github.com/JuliaMatric
es/BlockBandedMatrices.jl.git)
  [49dc2e85] Calculus v0.4.1
  [5ae59095] Colors v0.9.5
  [717857b8] DSP v0.5.2
  [a93c6f00] DataFrames v0.16.0
  [2b5f629d] DiffEqBase v4.31.2
  [225cb15b] DiffEqTutorials v0.0.0 #master (https://github.com/JuliaDiffEq/Diff
EqTutorials.jl.git)
  [0c46a032] DifferentialEquations v5.3.1
  [31c24e10] Distributions v0.16.4
  [38e38edf] GLM v1.0.2
  [28b8d3ca] GR v0.37.0
  [c91e804a] Gadfly v1.0.1
  [09f84164] HypothesisTests v0.8.0
  [7073ff75] IJulia v1.15.2
  [86fae568] ImageView v0.8.2+ #master (https://github.com/JuliaImages/ImageView
.jl.git)
  [916415d5] Images v0.17.0
  [c601a237] Interact v0.9.0
  [e5e0dc1b] Juno v0.5.3
  [b964fa9f] LaTeXStrings v1.0.3
  [093fc24a] LightGraphs v1.2.0
  [1dea7af3] OrdinaryDiffEq v4.19.0
  [3b7a836e] PGFPlots v3.0.1
  [91a5bcdd] Plots v0.22.5
  [438e738f] PyCall v1.18.5
  [d330b81b] PyPlot v2.7.0
  [6f49c342] RCall v0.13.0
  [ce6b1742] RDatasets v0.6.1
  [f2b01f46] Roots v0.7.4
  [6ac157d9] SatelliteToolbox v0.5.0
  [276daf66] SpecialFunctions v0.7.2
  [60ddc479] StatPlots v0.8.2
  [2913bbd2] StatsBase v0.27.0
  [b8865327] UnicodePlots v1.0.1
  [44dac47c] VMLS v0.2.0 #master (https://github.com/VMLS-book/VMLS.jl)
  [37e2e46d] LinearAlgebra
  [9a3f8284] Random
  [2f01184e] SparseArrays

(v1.0) pkg>

How to update the latest version without having to install each package again?

2 Likes

See Julia v1.1.0 has been released.

4 Likes

My apologies, if I was not clear. I am referring to that version: julia1.1.0. It is where I am presenting the problem stated above
Last version on my PC

julia> versioninfo()
Julia Version 1.1.0
Commit 80516ca202 (2019-01-21 21:24 UTC)
Platform Info:
  OS: Windows (i686-w64-mingw32)
  CPU: Intel(R) Core(TM) i3-4160 CPU @ 3.60GHz
  WORD_SIZE: 32
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, haswell)

Previous version, still installed, on my PC:

julia> versioninfo()
Julia Version 1.0.3
Commit 099e826241 (2018-12-18 01:34 UTC)
Platform Info:
  OS: Windows (i686-w64-mingw32)
  CPU: Intel(R) Core(TM) i3-4160 CPU @ 3.60GHz
  WORD_SIZE: 32
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.0 (ORCJIT, haswell)

julia>

Yes, the linked comment applies to exactly this scenario.

Continuing the discussion from Julia v1.1.0 has been released:

Leveraging all the experience built into the code base of rbenv … you could use jlenv.

Install:

  1. jlenv - this installs julia-build by default.

Then install Julia versions:

$ jlenv install v1.1.0

NOTE:
As further versions of Julia are released, the julia-build project gets updated, so you will need to update julia-build, before you can install a new version of julia.

1 Like

So I see a lot of options in this thread:

  1. jenv @taqtiqa-mark (cross-platform?)
  2. removing old install, moving unpacked folder and symlinking new like @oliver (cross-platform?)
  3. using the installer (dmg on OSX @RAbraham)

I did 3) and seemed to have quite some issues but I’m using Jupyter in combination. I added the new kernel afterwards and now after removing v.1.0 everything seems to be functional again.

I’m still left wondering what’s the ‘cleanest’ way to upgrade a complete installation including all packages? Looking forward to 1.3 so would love to know before the release.

bash possibly with some support for zsh and fish (I don’t use those and there aren’t any tests for them).
It may work on MacOS and Windows 10 with its linux subsystem but neither are used or tested by me.

Really dependent on community use/support so issue reports and PR’s welcome.

I should add that the ideal jlenv use case is multiple installs for a user, e.g. per-project.
The ideal 'julia-cookbook` use case is system wide installs.

Of course you can do as you wish (I use chef to manage my desktop with user scoped installs and no system wide julia), but those encourage reasonable/sound practice.

Linking other thread: How can I be able to use binary command "julia" in Mac OSX terminal?

Upgrading Julia in Ubuntu is such a pain for me. And it feels like it could just as well have been super easy…

Here’s what happens. When I do “sudo apt-get upgrade julia” it doesn’t upgrade to v.1.3 (which I expected), no instead it downgrades to v.1.0.4!

I have no idea how these things are maintained, but would presume that sudo apt-get… should work, right? Perhaps the task of upgrading the repository has fallen between the chairs or something?

Peter

The Julia people don’t maintain the Ubuntu repositories.

Moreover, the repository is pretty conservative and doesn’t give people the newest versions by design. You should use the official Linux binaries instead as suggested on the downloads page https://julialang.org/downloads/platform/ .

6 Likes

well, that’s because you’re using Ubuntu, and Ubuntu people wants ‘LTS’ software, maybe it’s time to change a distro,

https://www.archlinux.org/packages/community/x86_64/julia/

2 Likes

Could we make an Ubuntu package called get-julia which is just a bash script that wgets and unzips into the right location? That way, the script wouldn’t need to be upgraded, but re-running it would lead to the user getting an updated Julia version. With a little extra work, such a script could also presumably check for packages and install them as well. Thoughts?

1 Like

I don’t think that Ubuntu packages for installing software in user directories are encouraged. And it’s a lot of hassle. There are solutions like

instead.

FWIW, distribution releases (when the idea is applicable) are always a bit behind on software versions for all software. This is natural: everything in the release should be well-tested. It is understood that if you want the cutting edge, you install from sources, or use a PPA or similar.

1 Like

the jenv and julia-build project doesn’t seem to be maintained anymore, right? Latest version available vom julia-install is v1.2.0-rc1

Hi mks, and welcome to the Julia community.

I haven’t used jenv or similar projects, never really had a reason to, not sure what I’m missing out on.

There’s a similar project, outdated only worked on even older Julia, but from the discussion, it seems at least partially no longer needed, from the comment I directly link to:

1 Like