[] [] Don't try to remove the temporary dir on the first install

[NyanHelsing]

Purpose

Running docker-compose up mfr_requirements fails because it waits for user interaction. Removing -e lets pip install aiohttppretty without failing because pip doesn't know what to do.

Changes

Remove the -e flag from the requirements file

Side effects

This flag looks like its been here since the dependency was first added, but there is no reason stated. There may be a reason its there? If so we can look at --ignore-installed as an alternative

QA Notes

Probably no.

Deployment Notes

Probably no. This only affects dev environments.

[coveralls]

Coverage remained the same at 72.448% when pulling 6ba2bce on birdbrained:bf/fix-req-install into f943de8 on CenterForOpenScience:develop.

[coveralls]

Coverage remained the same at 72.448% when pulling 8c3f9c6 on birdbrained:bf/fix-req-install into f943de8 on CenterForOpenScience:develop.

[felliott]

The change is GTG, but please git commit --amend the commit message to follow the guidelines in https://chris.beams.io/posts/git-commit/ (The 50 char max in the subject isn't a hard requirements, but try to keep it <60 chars).

[NyanHelsing]

My bad, I thought my newline was putting it in the body not the subject. (Normally I don't use commit -m but an editor to write commit messages. Serves me right for trying something new.)

[felliott]

LGTM! Merging.

[cslzchen]

@felliott and @birdbrained

Removing -e mode on WB (and on MFR) causes improper installation of aiohttpretty. MFR passes because aiohttpretty is not used at all. When I try to import the package in the test, it fails as expected.

See comments here: CenterForOpenScience/waterbutler#321 (review)

Update: tests are fixed after upgrading aiohttpretty to 0.1.0.

See comments here: CenterForOpenScience/waterbutler#321 (review)

• TODO: we should remove or at least upgrade aiohttpretty for MFR