%time doesn't work for multiline statements

[asmeurer]

This is possibly related to #2293.

In [42]: 1 + \
   ....: 2
Out[42]: 3

In [43]: %time 1 + \
  File "<timed exec>", line 1
    1 + \
         ^
SyntaxError: unexpected EOF while parsing

In [47]: %time 1 \
CPU times: user 0.00 s, sys: 0.00 s, total: 0.00 s
Wall time: 0.00 s

In [48]: 1 \
   ....: + 2
Out[48]: 3

In [49]: %time 1 \
+ 2
CPU times: user 0.00 s, sys: 0.00 s, total: 0.00 s
Wall time: 0.00 s
Out[49]: 2

(to enter [49], I had to up arrow to [48] in the history and add %time before it).

In addition to the obvious issues here of it not working with multiline statements, in [49], notice that it just evaluates the last line.

[minrk]

Do any line magics support continuations? I wouldn't expect so.

If not, then perhaps this is perhaps better expressed as the more generic "line magics do not support continuations".

[fperez]

On Tue, Sep 4, 2012 at 11:57 AM, Min RK notifications@github.com wrote:

Do any line magics support continuations? I wouldn't expect so.

If not, then perhaps this is perhaps better expressed as the more generic
"line magics do not support continuations".

That's indeed the case, this is not specific to %timeit but generic to
all magics. We've never made any attempt to put in the logic necessary to
capture line continuations and apply the single-line filtering machinery to
the joined line. It's doable, but it hasn't been done.

I vote for closing this and opening an enhancement PR about line
continuations in general for all input flitering, which we can then resolve
as part of the big refactoring that @takluyver is undertaking on that front.

Note: I don't mean that Thomas has to fix this as well, simply that we can
handle it there, even if handling it means deferring it to a later cycle of
refinement of that refactoring. That decision will be up to Thomas, who's
tackling the brunt of that effort.

[takluyver]

Yep, see #2164 for the general case of this. I hope my refactoring can make this sort of thing possible, but I haven't figured out all the details yet (I was on a conference and holiday last week).

I'm closing this one, as I think #2164 already covers it.

[asmeurer]

Might the result of [49] be considered a separate bug here?

[takluyver]

At a glance, it's another consequence of the same bug - the line
continuation is ignored, so it sees +2 as a separate expression.