Switch to JSmol

[vbraun]

JSmol works now. The functionality of upstream's jsmol.php script is in worksheet_jsmol_data(). Also lots of bugfixes and cleanups.

[ppurka]

Do we need the development-bundle folder of jqueryui?

[vbraun]

Probably not, but that wasn't from my commit and I didn't touch it.

[ppurka]

Some comments:

1. That dev bundle folder can be deleted. I don't see any ill effects of removing that. Maybe @kcrisman knows how to not have it in the git history.
2. I think the icon for loading the jsmol needs to be both more prominent, and/or accompanied by a text saying 3D. It might even be better to have it like in jmol, - button saying something like "Load 3D plot". Currently it is a light blue icon which is barely visible and hardly conveys the fact that it loads the 3D plot.
3. When "Live 3D" is ticked in a worksheet, it does not take effect immediately. One needs to close the worksheet and reopen it. Edit: I think this is expected behavior given that the Typeset checkbox also does not update the contents immediately.
4. When "Live 3D" is ticked in a worksheet and we open the worksheet, a message comes up at the bottom saying "Application loaded". It should say "Loading 3D plots. Please wait" or something similar. On my machine it took a good full minute to load the jsmol. And it is not a very weak machine either (8G ram, core i7 ultrabook). I was testing on the latest version of Chrome and on Linux.

[vbraun]

The "Live 3D" checkbox means: any future applet is loaded immediately, like the typeset checkbox. This is IMHO the expected outcome from a checkbox (NOT: button), it only alters a preference but does not immediately change the view.

Feel free to draw a different icon, the one shown is the Jmol default.

[ppurka]

Yes, I already realized this point about the live 3d checkbox. I had edited my comment earlier.

About point 1: it will be incredibly fragile to remove the dev bundle from git history see https://help.github.com/articles/remove-sensitive-data/

[kcrisman]

My pull request to the branch isn't quite right yet, because it turns out that the jQuery dialog has a slightly different syntax than before (or, less likely, something else happened in Volker's commit about that) so the popout doesn't look quite right. Working on this.

[kcrisman]

Indeed, jQuery changed. http://jqueryui.com/upgrade-guide/1.10/#changed-title-option-from-html-to-text
After looking at the options, I think it's better to just add some CSS to make this right.

[kcrisman]

Hmm, we changed some CSS, but there are still all these original SCSS files. I am reluctant to double-do everything and would rather just get this figured out in plain CSS, but if we ever follow the instructions at https://github.com/sagemath/sagenb/blob/master/sass/readme.txt it will obliterate these changes. @ppurka - have you ever done that while following the "Miscellaneous Release Instructions"?

[kcrisman]

Okay, I have fixed that annoying CSS problem. If one of you can take a look at that and merge it into this pull request, then I think that all that remains is some heavy testing. I'm already happy with jsmol on a few browsers. Has anyone tried this in IE? Hopefully it would work well.

[gutow]

I have always updated both the css and the scss just in case. Then it does not matter how things are handled. I agree that the scss really does not make a lot of sense to me.

[kcrisman]

By the way, I propose that this change (and anything else we get in there) be labeled sagenb 0.11, to celebrate!

[kcrisman]

AAGH! I will never understand Git and all the extra crazy commits. I was sure I didn't have any extra stuff in this... but too late. This inability to remove history will get me every time, maybe I should have tried to cherry-pick. I will just go ahead and also merge my dumb css fixes now because I have thoroughly added too much to the history anyway. I'm sorry.

[kcrisman]

Okay, Jonathan, if you can test these changes, I think I am happy enough with it to have us get tested "for real" by merging it into sagenb proper, with the possible exception of ppurka's comments 2 and 4 at #241 (comment)
I think 4 might be hard to deal with because it seems to be in jmol itself, not in sagenb.

[gutow]

OK, I may have time this afternoon...As to #4 in ppurka's comment list. That is not going to change. The application is loaded, when that notice comes up. The remainder of the time is loading the data from the server and then rendering it. On a slow machine using Chrome, where things have to be processed into base64 ASCII by the server this is not going to be fast. This is especially true since we are now doing the base64 conversion in python which is interpreted. I do not see anything we can do. On my i3 netbook (no great shakes) running both the server and the browser, I do not see such a long delay. Wish I had a better answer, because I know waiting is annoying.

I will look into the possibility of another loading message; there is a way to call javascript from Jmol/JSmol so I might be able to have a message or spinner come up during the load. No promises on this. Thus this should not be a show stopper.

[kcrisman]

I agree, actually, though this may be an argument in favor of always auto-unchecking the "live 3d" box when reopening a worksheet... Volker's fix for the "sleeping" works fine, but I guess there is nothing one can do about it if ALL of them in a large worksheet need to open at once. On the whole, this is better than the current situation where a lot of people can't even use 3d plots.

[kcrisman]

And as to testing, keep in mind I just want to make sure I didn't break anything big accidentally with the CSS and cache busting. Feel free to test on Chromium, that should behave like Chrome, I would think (you should get the debugging messages in your terminal about Jsmol query etc. only on that browser, from my experience - still not sure why the jsmol/ isn't rendered on the others but instead one goes straight to the jsmol...).

[kcrisman]

Okay, Jonathan and I agreed this was ready for real-life testing. Thanks esp. to Volker for turning this into a working draft! @ppurka - if you see this, what goes into making a new package for http://trac.sagemath.org/ticket/16004 ?

[ppurka]

@kcrisman here is an outline of the steps: https://gist.github.com/ppurka/d10f668db113c2d1ddd4

It might also be better to access the gist from here: https://gist.github.com/ppurka in case I make any further changes.

[kcrisman]

Wow, that was really helpful - and I had forgotten about @kini's good work earlier on this. Hopefully I can just add some of these directly to the sagenb documentation if (when?) I try it.

[kcrisman]

See #250 - thanks again.