store_history is defaulting to False instead of True

[crew102]

Greetings. It looks like #1380 changed the default of store_history from True to False. This makes versions 6.0.2+ not backwards compatible with calls to nbconvert that used prior versions.

Incidentally, prior to version 6.0.2, the doctstring had indicated that the default was False (and not True), but it has always been True since that feature was first introduced in #1055.

[MSeal]

Two points:

One, the value of store_history should not impact actual execution of notebooks from a headless setting. It's only used to be able to fetch the history of commands when running in a live repl. There should be no compatibility impact -- where were you seeing such an issue care about this setting?

Two, store history was turned off because it A) polluted actual local history if one is also using a real kernel in the same environment and B) because it persisted notebook contents to disk and sometimes templates or runtimes were injecting code they didn't know was staying around on the machine after cleaning up the template files. The latter was more of a concern for papermill, which used to call nbconvert to execute and would sometimes inject secrets into parameter cells. That is not longer the case but the merits of not persisting by default stand imho.

[crew102]

The issue came about for me when I tried to fetch the history outside of an interactive interpreter (i.e., when running nbconvert). This used to work fine for me, but after updating to version 6.0.2, some functionality in my package (reprexpy) stopped working b/c the default was set to false.