Skip to content

Conversation

@aburgm
Copy link

@aburgm aburgm commented Sep 4, 2015

Is there a particular reason this is the other way around?

When the connection file is first written and then logged, then external programs could monitor the log output and use the connection file (e.g. connect a client) without race condition.

@Carreau
Copy link
Member

Carreau commented Sep 5, 2015

That's sill don't guaranties on all system that the log will be available after the connexion file.

Also ipykernel is the only one that (for historical reason) write it's own connection file. Your external program should write the connection file itself, and pass it to ipykernel as an argument.

But i'm ok with this change, though I don't want to guaranty and maintain that it will work.

minrk added a commit that referenced this pull request Sep 5, 2015
Write connection file before writing its location to the log
@minrk minrk merged commit eaf5d8b into ipython:master Sep 5, 2015
@minrk minrk added this to the 4.1 milestone Sep 5, 2015
@minrk
Copy link
Member

minrk commented Sep 5, 2015

This should be a fine change, but as @Carreau mentioned,

I really need to get to the jupyter kernel entrypoint, for providing a wrapper for launching kernels with a KernelManager always attached.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants