calculate arrow shaftScale using projection of ends onto shaft offset

[bergey]

I think this fixes the bug described in 56b703f#commitcomment-4294528. The output looks like this

[byorgey]

Ah, this is a much more sensible way to do it.

[byorgey]

Now we just need to fix the warning, and we can also do away with endTangent and startTangent, simply calling tangentAtStart and tangentAtEnd (from Diagrams.Tangent) directly on the Trail.

[jeffreyrosenbluth]

Puzzling. If you look closely at http://bergey.s3.amazonaws.com/diagrams/Arrowtest-fixed.svg you will notice that the tip of the orange arrow head is not in the exact center of the origin. This occurs in both @bergey 's fix and my alternative, but did not happen when I calculated the shaft scale iteratively (i.e. using binomial search)?

For comparison:

https://github.com/jeffreyrosenbluth/diagrams-gallery/blob/master/arrowtest.svg

[byorgey]

Are you sure it's not just a rendering artifact? Can you make a bigger version which clearly shows the discrepancy? It is too small for me to really see.

[jeffreyrosenbluth]

I'm not 100% sure but if you look at the diagram I made with old code you
can see that it's off.

Take a look, I put a link in the comment to an svg that runs the same test
using the iterative technique

On Thu, Oct 10, 2013 at 10:29 AM, Brent Yorgey notifications@github.comwrote:

Are you sure it's not just a rendering artifact? Can you make a bigger
version which clearly shows the discrepancy?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/125#issuecomment-26058300
.

[jeffreyrosenbluth]

I actually, just dropped my old shaftScale function (using fInverse) into
my newest lens code and the diagram renders correctly!

On Thu, Oct 10, 2013 at 10:31 AM, Jeffrey Rosenbluth <
jeffrey.rosenbluth@gmail.com> wrote:

I'm not 100% sure but if you look at the diagram I made with old code you
can see that it's off.

Take a look, I put a link in the comment to an svg that runs the same test
using the iterative technique

On Thu, Oct 10, 2013 at 10:29 AM, Brent Yorgey notifications@github.comwrote:

Are you sure it's not just a rendering artifact? Can you make a bigger
version which clearly shows the discrepancy?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/125#issuecomment-26058300
.

[byorgey]

I still cannot see the difference. How about this: can you (1) open a new ticket for this issue, and (2) make two very large (like, a width of at least 1000) images which show the discrepancy, preferably using a cut-down version of the diagram that includes only the first two circles and the arrow with the orange head.

[bergey]

Last night I spent some time in ghci comparing the outputs of various shaftScale functions. The project-and-divide version agrees with the fInverse version to 3 decimal places, with differences beyond that. I'm not sure why that is; I'm pretty sure that they're the same algebraicly.

(Incidentally, @jeffreyrosenbluth I agree that the code you pasted above is equivalent to the code I commited. I just wrote the one that occurred to me first.)

[jeffreyrosenbluth]

When I get a chance I will

1. Open a ticket
2. Try to create a more obvious example
3. Try to compare the actual svg generated.

Sent from my iPad

On Oct 10, 2013, at 10:48 AM, Daniel Bergey notifications@github.com wrote:

Last night I spent some time in ghci comparing the outputs of various shaftScale functions. The project-and-divide version agrees with the fInverse version to 3 decimal places, with differences beyond that. I'm not sure why that is; I'm pretty sure that they're the same algebraicly.

(Incidentally, @jeffreyrosenbluth I agree that the code you pasted above is equivalent to the code I commited. I just wrote the one that occurred to me first.)

—
Reply to this email directly or view it on GitHub.

[jeffreyrosenbluth]

1 and 2 are done.
I'm not sure if 3 is worthwhile - too many small rounding errors to find
the relevant info.
But I'll give it a try.

On Thu, Oct 10, 2013 at 11:04 AM, Jeffrey.rosenbluth <
jeffrey.rosenbluth@gmail.com> wrote:

When I get a chance I will

1. Open a ticket
2. Try to create a more obvious example
3. Try to compare the actual svg generated.

Sent from my iPad

On Oct 10, 2013, at 10:48 AM, Daniel Bergey notifications@github.com
wrote:

Last night I spent some time in ghci comparing the outputs of various
shaftScale functions. The project-and-divide version agrees with the
fInverse version to 3 decimal places, with differences beyond that. I'm not
sure why that is; I'm pretty sure that they're the same algebraicly.

(Incidentally, @jeffreyrosenbluth https://github.com/jeffreyrosenbluthI agree that the code you pasted above is equivalent to the code I
commited. I just wrote the one that occurred to me first.)

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/125#issuecomment-26059955
.