refactor and bugfixes for plot_directive - #3257

Cool. I think @dmcdougall will probably also want to review this, since he's been digging into this file pretty deeply in the past couple of days.

Looks good to me.

@matthew-brett Was there anywhere specific in the docs that didn't link to source? I'd like to do a local build to sanity check it works, and then I'll merge it.

I'm sorry, I'm ashamed to say I didn't test this other than on a doc I
was writing myself, with:

plot_include_source = True
plot_html_show_source_link = False

in the conf.py. I think you'll see the problem if you write a tiny
test rst file with plot directive with and without plots (one with
:nofigs:)...

On 7/15/14, Damon McDougall notifications@github.com wrote:

@matthew-brett Was there anywhere specific in the docs that didn't link to
source? I'd like to do a local build to sanity check it works, and then
I'll merge it.

---

Reply to this email directly or view it on GitHub:
#3257 (comment)

This doesn't break the links fixed in #3253; merging.

@tacaswell Trying to cherry-pick the merge commit into v1.4.x and get this error message:

$ git cherry-pick 674044e
error: Commit 674044e8f0f8c4c9ef13d6a5b5fa15e4d7fbc53f is a merge but no -m option was given.
fatal: cherry-pick failed

I'm happy to cherry-pick each commit, unless there's a procedure you'd like me take to address this.

Look at the parent list on the merge commit. Counting from 1 get the I'd of the main line (which is normally 1) and pass that to -m.

I think what is going on is that git could walk down either side of the loop to generate the diff, -m tells it which side is most like the other branch you want to pick too.

Cherry-picked in 88fc024

@tacaswell Thanks!