David Manthey
David Manthey
Does just changing change the `uglifyOptions` in `grunt_tasks/build.js` from `ecma: 6` to `ecma: 5` fix this?
I think we should see if the `uglifyOptions` just fixes this first. If not, declare it unsupported.
I just tried this out with latest master. I found two changes to `girder/web_client/grunt_tasks/build.js` were sufficient to build so we can load in IE11 without errors. (1) Add `babel-polyfill` to...
Part of the issue here is that searching for `test.png` will match `test` or `png`, so the result set is large. The user probably wanted to do a search for...
Batches won't fix the problem with exhausting memory for sorting. The sort has to happen before it can determine what to return in a batch.
I think if we are going to do it, we should write the *n*-hop aggregation, rather than just the two-hop. It probably would be a few hours of work, as...
I see that the coverage file sent to codecov on on a successful coverage branch from late August was [600kb](https://circleci.com/gh/girder/girder/12853), whereas a recent submission from the same step was only...
In the circleci environment, the last working coverage used pytest 3.7.3 and pytest-cov 2.5.1. The broken coverage has pytest 3.8.0 and pytest-cov 2.6.0. A crude test of pinning pytest
I see that working versions store coverage in a file called something like `girder/build/test/coverage/python_temp/.coverage.692a993cea2b.3978.292632`, whereas the newer version store coverage in `girder/build/test/coverage/python_temp/.coverage`.
I'll check if we have to pin pytest (I assume we have to pin pytest-cov). Pinning both definitely works.