the error and the fact that it works with an alias suggests that script-name is problematic
script-name comes from file-path->script-name, which is splitting the path on '/', which doesn't produce the desired result on a Windows canonicalized path
I haven't tested on a local jar yet, but I assume the same symptom and root cause would pop up there
I haven't started on a PR yet, but a little thinking about it raised another question in my head:
The current file-path->script-name takes the first split of the file name on .. Would it be desirable to drop the extension instead? By which I mean 'foo.bar.clj' would be installed as 'foo.bar'. I realize this is very unlikely to make a difference because of namespace names and what not, but I was considering using fs/strip-ext in the PR, which would technically be different in the edge case.
pre-emptive apology I don't mean to be the "but what about Windows?" guy, but...
problem statement Installing a local clj on Windows throws:
but works if an alias is supplied:
analysis
script-name
is problematicscript-name
comes fromfile-path->script-name
, which is splitting the path on '/', which doesn't produce the desired result on a Windows canonicalized pathI haven't started on a PR yet, but a little thinking about it raised another question in my head: The current
file-path->script-name
takes the first split of the file name on.
. Would it be desirable to drop the extension instead? By which I mean 'foo.bar.clj' would be installed as 'foo.bar'. I realize this is very unlikely to make a difference because of namespace names and what not, but I was considering usingfs/strip-ext
in the PR, which would technically be different in the edge case.