deactivate.bat doesn't cd back to %_LAST_DIR%

[thepacketgeek]

In cdvirtualenv.bat, there is logic setting %_LAST_DIR% to the current directory, but nothing in deactivate.bat to change back to the %_LAST_DIR% before activating.

Before I fork/MR, is this expected behavior for some reason?

Thanks!

[utek]

I'm ok with both versions. Not cd (as it is now) and cd to first dir as you propose. Maybe do it on option or some env variable (like VENV_CD_BACK or something). Anyway. Looking forward for PR if you'll think about something.

[thepacketgeek]

Great, I like your idea of having options as well. I'll work on this soon.

[thebjorn]

I'm a big -1 on this. It's surprising and annoying to be pulled from your flow, e.g.:

\> workon py27
(py27) \> cd my\source
(py27) \my\source> :: fix something / test something
(py27) \my\source> workon py35
(py35) \my\source> :: fix something / test something
(py35) \my\source> deactivate
\my\source> :: maybe run tox or work on docs in global sphinx..

if I got tossed back to \ after deactivate I would find it surprising and unwelcome. cd-.bat mostly does what is asked here - except for the %_LAST_DIR% being continually clobbered. We could fix this by either appending to %_LAST_DIR% instead of overwriting, or by using pushd and make cd- a synonym for popd.

[thebjorn]

deactivate.bat is not our file (although this might be changing -- see my comment on PR #73), making this difficult. This issue should stay open until our version of deactivate comes so we don't forget.