BUG: Fix an issue with the csv cparser when usecols is used

[tr11]

closes #3192

This pull request fixes a problem arising from using the csv parser with the usecols option. In that case, all the dtypes will be converted to object. This code fixes that issue, resulting in the correct dtypes.

Thanks,
Tiago

[tr11]

This pull request fixes issue #3192 .

[jreback]

that's what I like someone who fixes an issue they see!

• can you put an entry in release notes? (edit doc/source/release.rst)...follow the format
• can you turn on travis (so it builds on various confifgurations), see contributing.md for help

thanks!

[jreback]

looks like you need to rebase on master

git checkout master
git branch <this branch> --set-upstream master
git checkout <this branch>
git pull --rebase

fix any conflicts (which are most likely in release.rst)

also might want to rebase down to 1 commit

git rebase -i <commit before your first one>

then s or f

then

git push origin <this branch> -f

see here

https://github.com/pydata/pandas/wiki/Tips-&-Tricks

[tr11]

Should be ok now.

[jreback]

thanks!

[ghost]

This PR created a regression, #5766.