You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
How do we get to end goal of hitting a button to release on Github?
auto generate changelog, enforcing stricter syntax in PR description
figure out what backports should/could be done for each PR
How often do we release things?
If we can automate, it makes it less of a burden to actually do release
and so "when" it happens can be based more on first principles than on who is available (maybe?)
Should we be clearer about our intentions, both around getting PRs in and doing releases? So that we understand the conflicting priorities possibly between our employer's desires and the project's needs?
As a benchmark, this release took somewhere between 16-30 hours of Saul's time and 16-30 hours of Jason's time
Can we ship an extension that does not require a post-install phase that re-compiles the JupyterLab application? In other words, can an extension be a simple set of static assets that resides in a well-known location? Can I untar some file into a directory and refresh the page and have it magically work?
System is not magic for extension authors: Giving extension authors more control over the environment their code is built in, before being run in the browser, would be useful. For example, currently we run all extensions in our webpack config. This is an issue, if extensions have their own config they would rather run. It's also confusing for them, because the post install build step is opaque and not documented. So when it fails (as pointed out above) we get weird issues. Also, it fails on user machines, not extension athors machines, which is even harder to debug.
Track versioning between frontend extensions, backend extensions, jupyterlabs server versions, jupyterlab version. We currently have two package manager resolutions working seperately. It would be nice if we could just use one to properly resolve versions between frontend and backend code and extensions.
One package per extension, even if that involves both a frontend and backend component
A way to minimize the download size for clients on slow networkd
debuggability
Make it easier to develop multiple extensions independently
allow for the creation of seperately deployable built artifacts from each extension
The text was updated successfully, but these errors were encountered:
JupyterLab Weekly Meeting Minutes
08 Apr 2020 Weekly Meeting
Eric
Ed
Saul
Jason
Darian
Tim
Alex
Aditional Discussion:
The text was updated successfully, but these errors were encountered: