Allow file comments with genfromtxt(..., names=True) #351
Conversation
|
Ah. OK, so this change will not work with the following: (using the table from test_commented_header). But it will work with any of the following: etc. If this is an acceptable trade-off, I can rewrite the test. |
|
So it sounds like there's a behavioural change here, where before we required header names to have a comment marker at the beginning and now we require them not to? Or something like that? Discussions about what behaviour we want belong on numpy-discussion, because lots of people who might be affected don't read PRs, so you should send a note there explaining the issue. |
|
OK, will do. |
|
Discussion is at http://www.mail-archive.com/numpy-discussion@scipy.org/msg38306.html |
| if names is True and comments in first_line: | ||
| if skip_header == -1: | ||
| first_line = first_line.split(comments)[0] | ||
| else: | ||
| first_line = asbytes('').join(first_line.split(comments)[1:]) |
There was a problem hiding this comment.
FYI -- .split takes a second optional argument. All of these should just become first_line.split(comments, 1), and the join parts can just go away.
|
Ping the travisbot. |
|
I think this needs a test. @njsmith As you noted, this seems to change behavior. I don't recall the discussion about that or if anything was decided. |
|
A quick skim of that mailing list thread suggests that this change would break several people's code. I didn't read enough to see if a more generally acceptable solution was found. |
|
Sounds to me like it should be closed then. I don't know why travisbot it trying to install this in 2.4, but I've seen that in a few other spots. |
|
Going to close this. |
feat: Add vtst[q|d]_[s64|u64]
This fixes ticket #2184 (http://projects.scipy.org/numpy/ticket/2184).