Tcl module fixes #739
Tcl module fixes #739
Conversation
…est for dealing with a very long module path
…terse' output, and document it thoroughly
…ath runs first (required for Lmod)
…cmd.tcl with tclsh to make it robust against older versions of modulecmd.tcl that don't have a proper hashbang
…it for modulecmd.tcl
- run 'module use' on all module paths after changing $MODULEPATH - also set $LD_LIBRARY_PATH_modshare when (re)setting $LD_LIBRARY_PATH - also tweak stdout for purge/use module commands - deprecate passing of list of module paths to run_module; not used, not needed, only complicates things w.r.t. 'module use' after touching $MODULEPATH
|
Should be fully ready now, with all unit tests successful with all module tools we test (see https://jenkins1.ugent.be/view/EasyBuild%20(boegel)/, should be all blue). |
|
@stdweird: Some more issues popped up with the DEISA |
| # | ||
| 'available': re.compile(r""" | ||
| ^(?!-*\s) # disallow lines starting with (empty) list of '-'s followed by a space | ||
| \s* # ignore whitespace at start of the line |
There was a problem hiding this comment.
change comment, this line has nothing to do with start of line
There was a problem hiding this comment.
well, yes, it does, since the previous line is a lookahead...
There was a problem hiding this comment.
no, per the previous statement, the line can't start with a whitespace character
There was a problem hiding this comment.
hmmk, this is not what I intended here...
It's OK, because none of the modules tools we currently support include whitespace before module names with --terse, but it's more strict than I would like it to be.
I'll remove this part of the regex all together, since it's senseless.
| mods = super(EnvironmentModulesTcl, self).available(mod_name=mod_name) | ||
| # strip off slash at beginning, if it's there | ||
| # under certain circumstances, modulecmd.tcl (DEISA variant) spits out available modules like this | ||
| clean_mods = [] |
There was a problem hiding this comment.
you mean clean_mods = [mod.lstrip(os.path.sep) for mod in mods]
There was a problem hiding this comment.
ok, I should have realized that, fixed
|
@boegel reviewed |
…nd use which function for checking whether selected module command is available
|
|
||
| def which(cmd): | ||
| """Return (first) path in $PATH for specified command, or None if command is not found.""" | ||
| paths = os.environ.get('PATH', '').split(os.pathsep) |
There was a problem hiding this comment.
i would throw an error of some sort in case PATH is not available. won't find anything anyway
There was a problem hiding this comment.
The error message will show that no paths were considered ([]), isn't that sufficient?
There was a problem hiding this comment.
Enhanced the error message, it now clearly mentions $PATH
|
@boegel feel free to merge this in, we can chekc the shell issue later |
… that were already loaded
… module tools; stop setting $LOADEDMODULES empty if its not defined
… to modules.py; required for modulecmd.tcl which also keeps track of $MODULEPATH_modshare
…igger 'module use' on module paths
Various fixes w.r.t.
modules.pyand its unit tests.This should fix #733 and #734.
Currently still a WIP, since some tests still fail in certain circumstances, see https://jenkins1.ugent.be/view/EasyBuild%20(boegel)/