Skip to content

Conversation

yuvipanda
Copy link
Contributor

This is on top of the 4.2.1 release.

@yuvipanda
Copy link
Contributor Author

I'm running the js tests locally now to see if I Can repro the failures.

@minrk
Copy link
Member

minrk commented Jul 26, 2016

@yuvipanda #1633 should have fixed the travis builds on 4.x, if you want to rebase and see if it passes, now.

minrk added 26 commits July 26, 2016 08:12
jupyter#1214

There is no function to hide the header in edit mode. Now I know my syntax is wrong. I couldn't find a way to do this from the edit page javascript file. If I could get some guidance and than I could write it properly. Right now I just put the javascript within the html edit file.

Before:
...
This adds the missing method of the pager, and for which there is already a call line 96.

This is required for ipython/ipython#9304
…nfigurable

also make cookies httponly by default, since we do not need or want access to cookies in js
…xecutions happening before a kernel is set up

This is an attempt at taking care of jupyter#994. We implement two queues for cell executions. One is at the notebook level, to queue up executions when a kernel does not exist. Another is at the kernel level, to queue up messages when the kernel is not connected.

*Edit: We scrapped the notebook-level queue, and now just have a kernel-level message queue for when the kernel isn't connected*
fixes current test failures caused by bug in IPython 4.1rc1
…ssage

it's not really a random port, it's usually n+1
Closes jupyter#1040.

External JS files are loaded with `application/javascript` mimetype.

cc: @Carreau @michaelsbradleyjr
- fix some undefined aliases, tests for help output
- make uninstall symmetrical with install
…s_hidden

instead of actually listing because listdir can be slow.

closes ipython/ipython#9165
map is lazily evaluated, so doesn't show messages
…unknown language

Closes jupyter#1209

Are the tests in the correct file?
in case Python was built with an incomplete standard library (missing sqlite3)

we do the same thing in IPython history, etc.

closes jupyter#1307
Hi,

I believe I've found a bug in the way that OPTIONS requests are handled by the notebook server. Pre-flighted OPTIONS requests do not include authentication information, which means that we should not expect these to be authenticated and so should not be using the @web.authenticated decorator. Authentication will be validated when the real PUT/POST etc request is issued after the preflight.

FWIW this behaviour is documented by others in different contexts (see for example: http://giix.nl/2015/03/10/cross-origin-resource-sharing-cors-and-kerberos-webserver-auth/).
...
PR jupyter#1279 didn't auto-apply cleanly, so I had to do a manual backport.
…r during tests

mock jupyter environment variables during nb/server-extension tests
remove duplicate what's new bullets from the index page
…ssion via PATCH

- PATCH to kernel.id attaches to other kernel
- PATCH to kernel.name starts new kernel with given name

Previous kernel is shutdown after the change

...
This allows both kernel name and kernel id to be provided in the request, making the request implementation simpler.
default_url contains base_url, and so does connection_url

closes jupyter#1336
minrk and others added 14 commits July 26, 2016 08:12
When there are non-ascii characters in notebook file name, i.e., "文件名.ipynb", notebooks cannot be properly downloaded via configurable-http-proxy in [JupyterHub](https://github.com/jupyter/jupyterhub) and an HPE_INVALID_HEADER_TOKEN is reported. After I inspect the headers returned by the notebook, I find that the file name in Content-Disposition header is not escaped, which cause the error in [request module](https://github.com/request/request) of node.js. After escaping the file names, the bug is resolved.
This should avoid the issue several people have had with unstyled pages on misconfigured Windows systems. I think practicality beats purity here.

Closes jupytergh-1350
Please see the attached screenshot. It appears that hashes are unintentionally being shown in the text.

![screen shot 2016-04-15 at 3 51 37 pm](https://cloud.githubusercontent.com/assets/474290/14577047/c6379fcc-0323-11e6-882f-399953e54054.png)
Problem: When a client loses their websocket connection and then reconnects, this triggers the creation of new, identical zmq sockets with a collision in their routing IDENTITY. This causes messages to be lost, even though the websocket connection is established.

Fix: if a duplicate connection (same session id, same kernel id) is attempted, close the stale one.

My first thought was to refuse colliding connections, but for the most likely case of a flaky network where the earlier session hasn't been closed yet, this experience is quite unpleasant,
...
With this, I can release 4.2.1 this evening.
This is on top of the 4.2.1 release.
@yuvipanda
Copy link
Contributor Author

Seem to have snafu'd the rebase, I've re-done it in #1634

@yuvipanda yuvipanda closed this Jul 26, 2016
@Carreau Carreau modified the milestone: no action Aug 1, 2016
@github-actions github-actions bot locked as resolved and limited conversation to collaborators Apr 13, 2021
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants