Allow index_netcdf to reference relative paths#61
Conversation
|
Your changes LGTM. The failing tests are clearly because the test framework isn't freeing up database resources after tests:
The pattern of failures w.r.t. Python versions is I think arbitrary. |
Have ideas on how to address that? |
|
Yes. I've been improving my test code successively in various database projects. I'm not sure where this one is in the sequence, and I'll have to have a look around to find which repo has the latest "version", but I think I know what to do here and have existing code that does it somewhere. |
|
Cool. I have an illogically visceral reaction to merging a PR with failing tests (even if it's completely unrelated). So I'll hold off on the merge, and rebase after we get your test fixes. |
|
Turns out I encountered and fixed the problem in the other PR open on this repo. I'll address your comments and get that merged, then we can rebase our ongoing stuff onto that. |
A common use case for indexing a set of files is to cd into the local directory containing the files and then to run "index_netcdf *". This will fail, unfortunately, because the database tracks absolute paths and index_netcdf naively accepts the paths from the command line. The PR tells index_netcdf to use the normalized absolutized version of each path, so that the command line args will match up with the paths in the database.
jameshiebert
force-pushed
the
feature/abspath
branch
from
4ed705f to
ef50bed
Compare
|
After a rebase, all looks good. Merging. |
2 participants
A common use case for indexing a set of files is to cd into the local
directory containing the files and then to run "index_netcdf *". This
will fail, unfortunately, because the database tracks absolute paths
and
index_netcdfnaively accepts the paths from the command line.This PR tells
index_netcdfto use the normalized absolutized version ofeach path, so that the command line args will match up with the paths
in the database.