Is there really any reason to have a separate style for each component? Wouldn't that be significantly less efficient than a single style.min.css that's slightly larger, but only needs to be downloaded / cached once?

the vast majority of CSS is shared, so if we minify to separate style.min.css, we are essentially multiplying the amount of traffic / requests involved in fetching CSS by the number of pages, which is the opposite of the goal.

Agree with the 2 comments of @minrk .
we can have a per-directory this-component-style.less that define per-componenent local style, but it seem that
bundeling up everything in one file make more sens.

Also, this can be part of another PR, but could the login form be pure JS.
So you can actually catch a 403 and just prompt to re-login on the same page ?
Bonus point if we figure out a way to do it that also works with emacs client among other.

I have to re-dive into notebook-list and cluster-list, but something along the same would be great, to for example have a quick access list in the notebook file-open menu.

IPython.core.tests.test_display.test_image_filename_defaults uses the ipynb logo image, and needs to be updated for the new path.

@ellisonbg did you want to do the same separation of restful services and human on the js side that you did on the Python side?

Also, was I wrong that you were going to put the IPython stuff in a single 'ipython' tree, rather than unpacking each subcomponent at the highest level?

I do but we don't really have those cleanly separated in the js code right now.

Sent from my iPhone

On May 16, 2013, at 7:08 PM, Min RK [email protected] wrote:

@ellisonbg did you want to do the same separation of restful services and human on the js side that you did on the Python side?

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

I do but we don't really have those cleanly separated in the js code right now.

Okay, that's for another PR, then?

If #3321 has beed merged, shouldn't lots of those commits not even show up ?

Probably just needs a rebase for them to not show up, which should actually proceed cleanly since those commits are in master.

Short from my phone.

Are the Line number back in the right place?

Le mercredi 22 mai 2013, Brian E. Granger a écrit :

OK so I think I have found all known style problems. But there is a lot of
styles to check and I very well could have missed some. We might want to
make other changes to the org of the files, but let's start discussing this
version to work towards a merge.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/3325#issuecomment-18248107
.

Ahh, good catch I will do that this morning.

I think I already commented on te faulty commit. IIRC notebook.less
CodeMirror-lines , padding to 0.4 em should be set only to padding-top and
bottom, not left and right.

Le mercredi 22 mai 2013, Brian E. Granger a écrit :

Ahh, good catch I will do that this morning.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/3325#issuecomment-18288933
.

OK I have fixed line numbers. Only one other change I might make: create a services directory in static.

Here, for the record: this breaks @Carreau's little slideshow mode button, but that's non-official so I wouldn't worry about it. I figure Matthias can update it once the merge goes in.

Otherwise I can't see any problems after running a bunch of stuff.

Given we have no testing of JS, our only recourse is to put this in front of trained monkeys (aka our users and devs), and dog-food it as fast as possible. That will at least give us a fighting chance of catching any serious problems before 1.0.

My vote is to merge it now-ish.

Let me fix the custom.JS stuff first.

Sent from my iPhone

On May 23, 2013, at 4:52 PM, Fernando Perez [email protected] wrote:

Here, for the record: this breaks @Carreau's little slideshow mode button, but that's non-official so I wouldn't worry about it. I figure Matthias can update it once the merge goes in.

Otherwise I can't see any problems after running a bunch of stuff.

Given we have no testing of JS, our only recourse is to put this in front of trained monkeys (aka our users and devs), and dog-food it as fast as possible. That will at least give us a fighting chance of catching any serious problems before 1.0.

My vote is to merge it now-ish.

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

OK I have fixed the custom js/css stuff and have also styled the code/pre elements in rendered_html back to their pre bootstrap versions. As far as I know, this PR is ready to be merged. We can continue to test and fix the css styles in master - the main thing is that the re-org work doesn't seem to have broken anything.

FYI, I need to look at requie.js, but I intend to write a small ext where you say:
IPython.loadJsExt('foo') and it will look in static: for foo/main.js and inject it in notebook, so that you could have separated folder for plugins. So I indeed find that the user or custom folder is one level too deep for users. But it doesn't prevent us from advising using subdirectories...

Comments as I use it, I'll modify as long as I discover things.

• login form weird with button on new line.
• Prepare an hidden Username field ?
• Line number gutters rounded on top-right, shouldn't.
• stderr not any more in red ( 2 other styles overwrites background).
• heading-x got blue halo when clicked

FYI, I need to look at requie.js, but I intend to write a small ext where you say:
IPython.loadJsExt('foo') and it will look in static: for foo/main.js and inject it in notebook, so that you could have separated folder for plugins. So I indeed find that the user or custom folder is one level too deep for users. But it doesn't prevent us from advising using subdirectories...

This should probably be inside the custom dir, to clearly distinguish it as something not shipped with IPython. So loadJSExt('foo') would load static/custom/foo/main.js.

everything I am aware of has been addressed now.

Merging now...