h2 = apply(hip.p1, 1.0, hip.a1)

mirror((waist), [hip.p1:p2])

when

mirror((waist), hip)

would be simple and seems like a reasonable extension of the existing function.  I guess maybe not everyone would have the same notion of how reflection about path should be defined, as this doesn't preserve distances to the path.  Still when working with paths, this is something I find myself needing to do.  Especially when half-front and half-back of a garment are drafted on the same side of the body instead of mirrored.

If a path is curved, visually and psychologically the mirror effect will make little sense, I am afraid, as the reference (line as of now) should represent the fold line.

If, for example, you mirror back against a path that is straight but consists of two lines (e.g. center_back=path(upper_center_back, lower_center_back) where the two latter are lines), and mirroring against center_back returns an error, then you can simply mirror against upper_back. No need to use the whole 'fold line', you just need to define the vector, and a shorter line is enough for this.

But those are the sort of consideration the user already has to contend with--is the hip shape too severe to be useable?  In practice the most frequent way I'm using these functions is to compare intersecting seams at the endpoints of paths.  If I've drafted a shirt with the front and back concentric, I may want to check how my armholes match up even though my side seams aren't perfectly straight, or even piecewise straight.  I can use magnet to bring the two underarm points into alignment with respect to the side seam (and don't have to draw a temporary guide point to do it), but then I have to also reflect the back armhole with respect to the side seam at the underarm point, so that I can see the back armhole as a continuation of the front.  I then might make an adjustment to some or all of the four paths.  Then I undo the mirror and I undo the magnet (with respect to the original paths).  Then dispose of the old paths.  

I know how to do this without the added functionality to mirror, but it would make my thinking a little more consistent if mirror could handle an arbitrary path, too.  It also could be a somewhat silent functionality, that extends mirror but isn't emphasized in the manual.

E.g.

magnet((lower_armhole_back), side_seam_back, side_seam_front)

mirror((lower_armhole_back), side_seam_front){

The new function mirrors about a path by using the both endpoints

so,

mirror((something),path)

is just the old mirror function:

mirror((something),[path.p1:path.p2])

I don't really need that extension, because it's already possible to write with one line of code in an intuitive way.

What I want, is for mirror path to be a 'mirror' about the initial vector of the path.  Which right now, I would do like this, first constructing a temporary point p_:

p_ = apply(path.p1, 1.0, path.a1)

mirror((something),[path.p1:p_])

Which I don't like because it forces me to create a temporary line to mirror about when all the information is already encoded in the path.

I would also like to be able to mirror about the negative of a path, so that:

mirror((something),-path)

would be:

p_ = apply(path.p2, 1.0, path.a2)

mirror((something),[path.p2:p_])

If it's any help, here's a code fragment that suppresses length on a curve by adding a contour dart:

Lines 38 and 42 are 'magnet' being used with a path and a negative path, in an intuitive way, to close a dart.

Lines 51 and 59 are how I would like to use 'mirror' when truing the top edge of the dart.

Lines 52, 53 and 60, 61 are what I've done instead.

Anyway, that's my request.  I can use the existing mirror function to do what I want to do, so it's not essential to me.  I just find the way 'magnet' works to be extremely convenient, and constructing line segments to work with 'mirror' a little annoying.  And I thought it might be an intuitive construction for anyone who doesn't think to create a line segment.

Thanks, again!