wehi-researchcomputing / environmentmodules Goto Github PK

> module_avail('python')
 [1] "python/2.7.18"      "python/3.10.4"      "python/3.5.1"      
 [4] "python/3.5.3"       "python/3.6.5-intel" "python/3.7.0"      
 [7] "python/3.7.13"      "python/3.8.3"       "python/3.8.8"      
[10] "python/3.9.5" 

module_avail() doesn't show which one's the default Python. For example, in bash, the default module is underlined.

Automatically calculate the dependencies of a library's .so and dyn.load in reverse dependency order.

e.g. module_load("<TAB> should autocomplete do available modules.

At various points it makes sense to ask the user between a small number of options. We can do this using menu().

For example, if they module_load("python") it probably makes sense to ask them which version they want.

If they module_load("X") and there is a dependency collision, we might ask them which option they want: to swap the colliding module with the new module, to cancel, or to purge and then load the new module.

We no longer particularly need it since module_list and module_avail no longer output ANSI strings.

 > module_load("python/3")
✔ Successfully loaded python/3

> module_list()
[1] "python/3.8.8"
> module_load("python/2")
Loading python/2.7.18
  ERROR: Conflicting python is loaded
✔ Successfully loaded python/2

The successfully loaded ... should be dropped or updated to provide correct info.

--json is supported since 4.5. If this is available, we should use this to provide a more native R output, such as a character vector of modules when using module_avail().