Are we even able to assume that MultipleLocator deals only with numerical values?

I think that if you tried to put any other type in it would be cause issues going through the Base class.

This seems worthy of backporting to 1.2.x.

re-based to v1.2.x

I really wouldn't do this. I'll admit that I'm not sure what the right fix is yet, but this particular fix gives very little control over when the last tick appears and when it doesn't. What if I'm plotting things at very small scale, such that the 1e-15 matters?

I'll go digging, but there has to be some code somewhere that is doing the chopping of the locator "locs" to the current view limits. It's that code that needs to have a proper tolerance implemented, if it doesn't already.

@dhyams That is an extremely good point and should be the reason not to accept this PR.

I have submitted another pull request to address this problem, as an alternative to this one.

#1686

I feel that #1686 poses a slightly more appropriate solution so I'm going to close this PR in favour of #1686.

If @tacaswell (or anybody else) feels that this PR is indeed preferred over #1686, please feel free to re-open this.

I hope my feelings about this patch is clear from the name of the PR.

On Sun, Jan 27, 2013 at 1:00 PM, Damon McDougall
notifications@github.comwrote:

I feel that #1686 https://github.com/matplotlib/matplotlib/issues/1686poses a slightly more appropriate solution so I'm going to close this PR in
favour of #1686 #1686.

If @tacaswell https://github.com/tacaswell (or anybody else) feels that
this PR is indeed preferred over #1686#1686,
please feel free to re-open this.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/1659#issuecomment-12759174.

Thomas Caswell
tcaswell@gmail.com