!SDKYnTgYyOXNBpJySp:matrix.org

cylc protocols & APIs

18 Members
HTTPS, WebSocket, ZeroMQ; REST, GraphQL1 Servers

Load older messages


SenderMessageTime
11 May 2021
@kinow:matrix.orgBruno P. Kinoshita I have the commands I used (should be bash, with curl...) buried somewhere... 11:03:02
@oliver-sanders2:matrix.orgoliver sandersAny thoughts on https://github.com/cylc/cylc-uiserver/issues/219? Not sure how easy/hard this would be and how nicely we could expose the Cylc endpoints to other users.11:05:03
@kinow:matrix.orgBruno P. KinoshitaI only read issue description, but looks like a good idea. I had a bookmark for... jupyter-proxy-server? Another way of hosting apps in Jupyter I think... There's a researcher at NIWA, Maxime, who did the R Shiny integration for JupyterHub recently. He showed me how he added a button in JupyterLab that launches R Shiny (it starts it the first time you click the button if R Shiny is not running, but only runs 1 instance). I tried doing something similar for Cylc UIS but that didn't work šŸ¤£ I think they must be using something like jupyter-server for R Shiny at NIWA already?11:08:54
@kinow:matrix.orgBruno P. Kinoshita(found the bookmark by typing jupyter proxy in the URL bar: https://jupyter-server-proxy.readthedocs.io/en/latest/server-process.html)11:09:50
@kinow:matrix.orgBruno P. Kinoshita(also found Maxime's repository: https://github.com/nesi/rstudio_on_nesi, my idea was to have a Cylc logo like the R logo in his screenshot; šŸ›Œ time šŸ‘‹)11:48:35
@oliver-sanders2:matrix.orgoliver sandersDid some quick experimentation, was able to host the static resources and graphql endpoints on a Jupyter server.16:19:40
@oliver-sanders2:matrix.orgoliver sanders We would end up tidying our Cylc endpoints behind a url path e.g. cylc/ so the UI would be at <uis|hub/users/<user>>/cylc/#/ 16:20:57
@kinow:matrix.orgBruno P. KinoshitaWe might be able to change it before releasing 8.0.0 then šŸ‘ 18:27:58
@dwsutherland:matrix.orgDavid SutherlandSort of change you want to do now, before investing too much more time in the hub (?)19:26:40
@dwsutherland:matrix.orgDavid Sutherlandunless Iā€™m missing something (servers replace notebooks, and lab replaces hub?)19:29:51
@davidmatthews:matrix.orgDavid Matthews

Yes, we're thinking we should make this change soon if we're going to.
Hopefully quite a small change but with lots of potential benefits.

(servers replace notebooks, and lab replaces hub?)

No - the hub remains unchanged.
The UIS becomes a jupyter server extension.
Rather than spawning the UIS we spawn jupyter server which runs the UIS as an extension.

20:11:26
@davidmatthews:matrix.orgDavid Matthews *

Yes, we're thinking we should make this change soon if we're going.
Hopefully quite a small change but with lots of potential benefits.

(servers replace notebooks, and lab replaces hub?)

No - the hub remains unchanged.
The UIS becomes a jupyter server extension.
Rather than spawning the UIS we spawn jupyter server which runs the UIS as an extension.

20:11:53
@davidmatthews:matrix.orgDavid Matthews *

Yes, we're thinking we should make this change soon if we're going to.
Hopefully quite a small change but with lots of potential benefits.

(servers replace notebooks, and lab replaces hub?)

No - the hub remains unchanged.
The UIS becomes a jupyter server extension.
Rather than spawning the UIS we spawn jupyter server which runs the UIS as an extension.

20:13:15
@dwsutherland:matrix.orgDavid SutherlandOh right šŸ‘20:15:11
12 May 2021
@dwsutherland:matrix.orgDavid Sutherland Bruno P. Kinoshita: found the place you get that extra delta after the workflow is shutdown/pruned 02:44:58
@kinow:matrix.orgBruno P. KinoshitaReally?! 02:45:19
@dwsutherland:matrix.orgDavid SutherlandCan't remember why I put it there02:45:23
@dwsutherland:matrix.orgDavid SutherlandYeah02:45:27
@dwsutherland:matrix.orgDavid SutherlandI think it was as some kind of signal02:45:46
@dwsutherland:matrix.orgDavid Sutherlandglad I'm actually spending time on this..02:47:55
@kinow:matrix.orgBruno P. Kinoshita Will be happy to review/test it if you prepare a PR David Sutherland . And I'm glad as well :-) 02:49:02
@dwsutherland:matrix.orgDavid SutherlandWell, it's all kind of related to getting a clean pruned signal through02:49:55
@dwsutherland:matrix.orgDavid SutherlandSo I just have to hunt down the issues you've (and others) have created to link the closure02:50:38
@dwsutherland:matrix.orgDavid Sutherland I guess the UIS doesn't need the shutdown topic anymore (put it in before scan, so the scheduler could signal it's last dying breath and the UIS/other can know about it) 03:27:22
@dwsutherland:matrix.orgDavid Sutherland * I guess the UIS doesn't need the shutdown topic anymore (put it in before scan, so the scheduler could signal it's last dying breath and the UIS/other can know about it) 03:28:14
@dwsutherland:matrix.orgDavid SutherlandUnless we still think that information is important to the UIS.. Will leave it in..03:33:32
@revilo666:matrix.orgHilary OliverNot sure, off hand, that might need some thought or discussion.03:36:33
@dwsutherland:matrix.orgDavid Sutherland at the moment the UIS grabs that shutdown signal and creates a delta with just delta.id and delta.shutdown in it. 03:48:26
@revilo666:matrix.orgHilary OliverI guess leave it in for now, but we should remove it if we can't think of a reason to use it.04:43:46
@dwsutherland:matrix.orgDavid SutherlandPRs updated (left the shutdown delta in)05:16:11

There are no newer messages yet.


Back to Room List