Closed pmspire closed 1 year ago
Hey @pmspire, thanks for reporting! Yes that makes sense. Can you open a PR that fixes the issue for you? I will make sure it is merged quickly.
Will do ASAP, thats @whilo .
I have a provisional fix, but am so far unable to test it, because
CompilerException
mentioned above, which I do not understand because key-vec
is let
-bound on the previous line. I tried replacing key-vec
with nil
in the problematic expression, but then getclojure.lang.Compiler$CompilerException: java.lang.IllegalAccessError: new-fs-store does not exist
, which seems to be true -- new-fs-store
is defined in filestore.clj
on master
, but not on development
.I would be happy to open a PR with my untested code, or to try any suggestions you might have.
Sure, feel free to open a PR, I just don't have nfs around atm.
Should be closed with the PR.
I am using
fs-store
on an NFS volume, and am getting (partial) stacktraces similar toIn
filestore.clj
, it looks likelist-files
does no filtering, unlike for examplecount-konserve-keys
, which filters for.ksv
files. I wonder iflist-files
may be seeing temporary.nfs*
files (in case you're not familiar: http://nfs.sourceforge.net/#faq_d2) that, by the time they are later referenced, no longer exist; and whether perhaps filtering out fileskonserve
should not care about (if that's possible to define) earlier might avoid this.I am working with
0.6.0-SNAPSHOT
due to issue #60 opened by my colleague. (Specifically,0.6.0-20211004.084316-6
-- I tried the latest0.6.0-SNAPSHOT
and gotclojure.lang.Compiler$CompilerException: java.lang.RuntimeException: Unable to resolve symbol: key-vec in this context, compiling:(konserve/filestore.clj:640:25)
, just FYI.)Thanks in advance for any thoughts you might have.