search

cortner / JuLIPMaterials.jl

Collection of Tools for Computational Materials Science
Other
9 stars 3 forks source link

Bug when freshly using module #14

Open lifelemons opened 6 years ago

lifelemons commented 6 years ago

When cloning and using the repo for the first time the error ERROR: LoadError: LoadError: UndefVarError: cluster not defined appears, while loading src/BCC.jl. After the second call of using JuLIPMaterials it loads fine.

Code and comments below, running in an environment that doesn't have JuLIPMaterials

using JuLIPMaterials # package not found error, as expected
Pkg.clone("https://github.com/cortner/JuLIPMaterials.jl.git") # clones fine
using JuLIPMaterials # error cluster not defined
using JuLIPMaterials  # no errors, seems to load fine

Not sure what cluster is as if one searches for it, it gets export-ed in src/common.jl, but it is not defined in there (nor in JuLIP or StaticArrays which are used in src/common.jl)

cortner commented 6 years ago

which branch of JuLIP and which branch of JuLIPMaterials are you on?

lifelemons commented 6 years ago

Occurs for both JuLIP:v0.2.2 and JuLIP:master and when currently cloning JuLIPMaterials it clones the master branch.

cortner commented 6 years ago

ouch, this is a big mess that I created here . . . I'll need to think about it. Sorry about this.

cortner commented 6 years ago

can you tell me quickly what you are trying to do? are you at all open to switch to the new JuLIP (branch jatoms)? Would probably take you 1-2 hours to convert your codes - I can help with that.

lifelemons commented 6 years ago

No worries. In terms of using JuLIPMaterials, I'm barely using it, don't use cluster or even src/BCC.jl. I just noticed this issue and thought I would mention it. Its relatively easy for me to navigate around it.

cortner commented 6 years ago

the problem is that the master branch assumes that you have the jatoms branch of JuLIP. So I'll need to create a new branch of JuLIPMaterials that uses JuLIP v0.2.2 but is also renamed (and not MaterialsScienceTools anymore)

lifelemons commented 6 years ago

Ah. In my case, so I don't use JuLIP:v0.2.2, though Im guessing the same solution/branch would work for the JuLIP:master branch. If that is the solution, I wouldn't worry about it, seems messy, and as I'm guessing the ultimate goal is to merge jatoms -> master, this issue eventually dissappears. (This issue is not a hinderence to my work so Im in no rush for a solution) Up to you on how/if you want to tackle it/solve the intermediate stage while jatoms is not mergered.

cortner commented 6 years ago

in principle you could just checkout the jatoms branch now, I'm really just waiting for James to "ok" the merge. It is fully functional. Are you using ASE (Python) at all?