TypeParameterAccessors not found error on Julia v 1.10.0

[andrewkhardy]

I have managed to break a Julia v.10.0 project install.

When I attempt to install ITensor on a HPCC, I get the following error

pkg> add ITensors
   Resolving package versions...
    Updating `/gpfs/fs1/home/a/aparamek/andykh/*my_project*/Project.toml`
  [9136182c] + ITensors v0.8.0
    Updating `/gpfs/fs1/home/a/aparamek/andykh/*my_project*/Manifest.toml`
ERROR: KeyError: key "TypeParameterAccessors" not found
Stacktrace:
  [1] getindex
    @ Base ./dict.jl:498 [inlined]

I have tried deleting the Project and Manifest files, the entire .julia directory, the entire julia-depot-x86_64 directory, all to no avail. Does anyone have suggestions for being able to start a project over from scratch? The install works on a version 1.9.3 I have not messed with, and this cluster does not have v1.11 or a way to run juliaup on the compute clusters, so I’m left trying to fix or clear my broken ITensor installation.

[mtfishman]

Please see the latest NEWS item: Introduction · ITensors.jl, unfortunately installing the latest version of ITensors.jl is broken on older releases of Julia v1.10 and v1.11, try using the most recent versions (Julia v1.10.8 or v1.11.3). That is good practice anyway.

EDIT: The NEWS item was updated to reflect that this issue has been fixed, please see my post below.

[mtfishman]

If you can’t update your version of Julia, you could try ITensors v0.7, though I would recommend using the latest versions of Julia and ITensors.jl if at all possible.

[andrewkhardy]

Thanks for the quick response Matt. This cluster admin is quite conservative on getting new modules. I will work to get juliaup working on their compute nodes and will make sure to read the News section more regularly in the future!

Thanks again.

[mtfishman]

This issue was fixed in [NDTensors] Support for older versions of Julia 1.10 and 1.11 by mtfishman · Pull Request #1623 · ITensor/ITensors.jl · GitHub, so when you upgrade to NDTensors v0.4.4 you can use the latest version of ITensors.jl with any Julia v1.10 or v1.11 releases.

However, I would still recommend using the latest releases of Julia if you are able to, since they come with bug fixes, performance improvements, etc.