I think I'll hold out for an intrinsic notation, like x[k,\cdot,\cdot], or
x[k,\vdots,\vdots].
Or x[k,_,_]
I'd enjoy writing array manipulations more without squeeze's interspersed.
On Wednesday, September 3, 2014 1:51:13 PM UTC-5, Tim Holy wrote:
>
> Sure, if you want to be reckless you can put this into your .juliarc.jl:
>
> import Base: squeeze
> squeeze(A::AbstractArray) = squeeze(A, filter(i->size(A,i)==1,
> 1:ndims(A)))
>
> You can of course choose to submit the second line as a pull request, and
> see
> what folks think.
>
> --Tim
>
> On Wednesday, September 03, 2014 11:35:48 AM David Smith wrote:
> > Ok, so you can continue using the old squeeze. Us reckless types can
> use
> > the aggressive one. ;-)
> >
> > I don't see why it shouldn't be available. Is there a fear that new
> users
> > will run into subtle errors and hate Julia because of it?
> >
> > What I hate is typing squeeze(x, 1) all the time.
> >
> > On Wednesday, September 3, 2014 1:21:56 PM UTC-5, Tim Holy wrote:
> > > On Wednesday, September 03, 2014 10:57:16 AM David Smith wrote:
> > > > Also, why don't we have a squeeze(A::AbstractArray{T,N}) method that
> > > > eliminates all singleton dimensions? If we had that, indexing
> returning
> > > > singleton dimensions wouldn't be as bad. Currently, having to write
> > > > squeeze(a[k,:,:],1) is terrible. That is something where Matlab
> wins.
> > >
> > > It wins, until the day that the a that gets returned by some other
> > > algorithm
> > > happens, on one iteration out of 10^4, to have size (K, M, N) with M
> =1.
> > > At
> > > that point, Matlab's squeeze delivers you an Nx1 array rather than the
> 1xN
> > > array you wished you had.
> > >
> > > I can't tell you how many times I got bit by that, until I stopped
> using
> > > squeeze altogether and just used reshape(). Which is more tedious yet
> than
> > > Julia's squeeze().
> > >
> > > --Tim
>
>
