BUG/TST: Allow generators in DataFrame.from_records

[tinproject]

closes #4910

• nrows implementation doesn't allow unknown size iterator like
  generators, if nrows = none ends with a TypeError.
• To allow generators if nrows=None consume it into a list.
• Add two tests of generators input.

[jreback]

pls add release notes (doc/source/release.rst), referencing the issue
pls hook up travis (see contributing.md)

thanks!

[tinproject]

Sorry for the mess, I'm new to github.
I've tested it locally with python 2.7 and 3.3 but didn't work with 2.6. Now it's all ok.

[jtratner]

I recognize that this may be silly, but how is this supposed to work if you passed something like this:

gen = ('a' for _ in range(50))
DataFrame.from_records(gen)

Just wondering if this needs to be caught anywhere or if this is okay

[tinproject]

I suppouse it's ok, the patch only put the generator in a list in the same fashion as main constructor.
DataFrame.from_records() expects a collection of records, which every record it's a collection of fields: list, tuple, dict, or in your case a string wich it's a sequence (collection) of characters (only one character in your example).

[jtratner]

it's fine, it looks like it just works the same as a list. Doesn't need to be added as a test case either.

[jreback]

@tinproject can you rebase and squash to commit? otherwise looks ok

[tinproject]

Hope it's ok, if it's not please give me a git command example, as i said i'm new to git :(

[jreback]

pretty good instructions here: https://github.com/pydata/pandas/wiki/Using-Git, let us know

[tinproject]

I think it's now ok. Release notes file release.rst conflicts but are trivial and I thought I solved it.

[jreback]

try git rebase -i upstream/master need to remove that merge of master commit

[tinproject]

@jreback now it's all ok, no? I must improve on git, sorry.

[jreback]

yes...looks good.....let me review

[jreback]

git just comes from practice! submit more PR's!

[jreback]

@tinproject thank you!