Release 5.5

[gnestor]

To do:

• Add 5.5.0 to changelog: Add changelog for Notebook 5.5 #3590
• Release 5.5.0rc1
  • Upload to PyPI
  • Post to Google Group
• Release 5.5.0
  • Create release on Github: https://github.com/jupyter/notebook/releases/tag/5.5.0
  • Upgrade notebook version in conda-forge/notebook-feedstock
  • Post to Google Group
  • Publish blog post

[gnestor]

@takluyver Can I help with publishing releases, announcements, etc?

[takluyver]

Please do! I've put the RC out on PyPI and emailed the jupyter list to advertise it. I think we should give that a few days before doing the final release, but if you can think of any other ways to get more people to try the RC, that would be welcome.

Brian also mentioned that he'd seen an issue with leaking file descriptors, but I don't even know how to monitor that, so any investigation is welcome.

[gnestor]

Brian also mentioned that he'd seen an issue with leaking file descriptors, but I don't even know how to monitor that, so any investigation is welcome.

@ellisonbg How can I reproduce this?

[kevin-bates]

I'm interested in the FD leak as well. Although not directly related to FDs, there were a couple jupyter_client PRs (360, 361 - referenced in #3424) that might be worth grabbing to see if this still can be reproduced. That said, since this appears to be new, I doubt there's a relationship.

[takluyver]

@kevin-bates do you know how to monitor an FD leak? If you can try to investigate that would be valuable.

[ellisonbg]

I have only seen it once. If this is the issue, running the notebook server, and opening, closing lots of notebooks/kernels would help to re-created it. I don't remember the best way to track down the source of FD leaks though. Min may have ideas.

…

On Thu, May 3, 2018 at 11:20 AM, Thomas Kluyver ***@***.***> wrote: @kevin-bates <https://github.com/kevin-bates> do you know how to monitor an FD leak? If you can try to investigate that would be valuable. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#3591 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AABr0Fyc2PucIXyfrthgC7XiOxAGCG6aks5tu0qCgaJpZM4Tvs80> .

-- Brian E. Granger Associate Professor of Physics and Data Science Cal Poly State University, San Luis Obispo @ellisonbg on Twitter and GitHub [email protected] and [email protected]

[kevin-bates]

For my testing, I primarily focused on memory leaks (using instrumentation) but also monitored socket (fd) usage via: ls -l /proc/<jupyter pid>/fd | grep socket | wc -l during kernel start/stop sequences (although I didn't [and don't] see anything wrt fd leaks). For a single kernel I see the socket counts consistent at 23 and 7 depending on whether its running or not.

[takluyver]

I played around with starting and stopping kernels while monitoring /proc/.../fd for the notebook server, and I couldn't reproduce an obvious leak. The count seemed to stabilise at 24 fds. So I think we can leave that until someone can clearly reproduce it.

[takluyver]

5.5.0 is now out on PyPI. I'll wait to see if the conda-forge auto update bot handles the update there.

[gnestor]

@takluyver Do you want to wait until conda-forge is updated before announcing on Google Group or blog?

[ellisonbg]

Thomas, thanks for investigating the fd issue and moving forward with the release!

…

On Wed, May 9, 2018 at 10:54 AM, Grant Nestor ***@***.***> wrote: @takluyver <https://github.com/takluyver> Do you want to wait until confa-forge is updated before announcing on Google Group or blog? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#3591 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AABr0MUZz6tmCKBM7z7KyLxmU8ZJfwa9ks5twy1BgaJpZM4Tvs80> .

-- Brian E. Granger Associate Professor of Physics and Data Science Cal Poly State University, San Luis Obispo @ellisonbg on Twitter and GitHub [email protected] and [email protected]

[gnestor]

I have created a draft blog post: https://blog.jupyter.org/jupyter-notebook-5-5-0-17b2e2d4f7f

Anyone care to review/publish?

[damianavila]

Anyone care to review/publish?

LGTM, but I do not have permissions to publish it.

[takluyver]

The first bullet point is a bit confusing:

Download as… menu item for nbconvert exporters

There already were menu items for nbconvert exporters. But whereas we previously showed only a fixed list, we now add menu entries for any exporters that have the export_from_notebook attribute set to a string. That's trickier to explain; I might take it out of the headline features list.

Other than that, feel free to publish it whenever you're ready.

[takluyver]

I see it was published, so closing this issue. Thanks @gnestor :-)

[gnestor]

I updated with "Dynamic Download as… menu for nbconvert exporters." @takluyver feel free edit if necessary 👍

[takluyver]

I don't think I have access to the blog - and I'm fine with this, there's plenty of things to do already.