Discovery on nested dtype numpy arrays and cleanup

[cpcloud]

• record dshape now prints using pprint
• discover now works for nested dtype numpy arrays
• removes some functions that are not used

[cpcloud]

@mrocklin any comments here?

most important commits are:

• https://github.com/cpcloud/datashape/commit/67ec34ef33a8eb3c3decef0af9ae07f81b0892a2
• https://github.com/cpcloud/datashape/commit/4e3936cbd352784c008b167e1410ad52ea1ba9df

[cpcloud]

i tried to be a bit more surgical with my commits because of the "cleanup" commits

[mrocklin]

Seems fine to me. My brain is mostly jello right now though.

Regarding commit practices (totally unrelated to this PR): Looking at the commits you mentioned made me wonder if its better to keep tests and code in the same commit so that future readers don't have to cross-reference.

I've recently become very lazy in my git practices though, so I'm not really in a position to make judgments.

[cpcloud]

if its better to keep tests and code in the same commit so that future readers don't have to cross-reference.

ah, yes. the reason i don't do it that way, is so that i can

1. commit the test
2. run it to make sure it fails
3. write the code that fixes the bug
4. test now passes

otherwise i have to stash and i've lost work that way in the past

[mrocklin]

Hrm, I would do the same thing, but replacing the word commit with write in the first line, I would then commit at the end. I probably don't have the same setup as you do though or something.

[cpcloud]

sometimes i write the code first and i don't want to lose what i've done, left that detail out :)

[mrocklin]

Alrighty

On Thu, Jan 8, 2015 at 1:42 PM, Phillip Cloud notifications@github.com wrote:

sometimes i write the code first, left that detail out :)

— Reply to this email directly or view it on GitHub https://github.com/ContinuumIO/datashape/pull/124#issuecomment-69254606.