Implement FPS-based Voila server

[davidbrochart]

References

Closes #977

Code changes

• Add a Voila fps flag, allowing to use FPS instead of Tornado.
• Have VoilaHandler depend on a server-agnostic _VoilaHandler class, so that the FPS server can reuse most of the handler logic.
• Create a fps-voila FPS plugin, which implements a FPSVoilaHandler based on _VoilaHandler.

User-facing changes

By default, Voila is backed by jupyter-server (no change). To use FPS, launch Voila with the --fps flag.

Limitations

This is still a POC, which allows to serve a notebook with widgets, but:

• no progressive HTML rendering,
• no tree view,
• and probably more broken functionalities!

[github-actions]

👈 Try it on binder (branch davidbrochart/voila/jupyverse)

[trungleduc]

@davidbrochart #936 is merged, NotebookRenderer is now available.

[davidbrochart]

Thanks, I will look at it.

[davidbrochart]

I just rebased, now that #936 is merged.

[trungleduc]

Hi @davidbrochart, I know this PR is still in the early phase but what do you think about the idea of refactoring the current Voila class into 2 parts, one is the CLI application to deal with the configurations, initialization... and the second part is the webserver application. So that we can have two interchangable servers implemented with tornado and fps.

[davidbrochart]

That sounds good @trungleduc, but what do you mean exactly? Having two methods, one for launching the Tornado application and one for launching the Uvicorn application?

[trungleduc]

I'm thinking about having a common interface for both and the CLI app does not need to know about the webserver implementation.

[davidbrochart]

But then how do you switch between jupyter-server and FPS? Currently it is passed as a CLI option:

voila my_notebook.ipynb --fps

[trungleduc]

Maybe we need to think more about it but the first thing popped in my head is to keep a mapping between the server configuration (--fps, --jupyter_server, ...) and the class of server implementation, then the CLI app can instantiate the selected web server with self._server_class_mapping[self.server_config](args...)

[github-actions]

Benchmark report

The execution time (in milliseconds) are grouped by test file, test type and browser.
For each case, the following values are computed: min <- [1st quartile - median - 3rd quartile] -> max.

Results table

Test file	basics.ipynb	bqplot.ipynb	dashboard.ipynb	gridspecLayout.ipynb	interactive.ipynb	ipympl.ipynb	ipyvolume.ipynb	multiple_widgets.ipynb	query-strings.ipynb
Render
chromium
actual	3010 <- [3020 - 3156 - 3176] -> 3520	2912 <- [2916 - 2938 - 3006] -> 3093	3014 <- [3121 - 3203 - 3283] -> 3418	3412 <- [3818 - 3821 - 3869] -> 3894	2037 <- [2038 - 2047 - 2071] -> 2124	3974 <- [4060 - 4229 - 4513] -> 5782	14836 <- [17250 - 18231 - 19158] -> 19177	12668 <- [12704 - 12836 - 12951] -> 13236	1629 <- [1684 - 1698 - 1771] -> 1915
expected	3379 <- [3442 - 3517 - 3701] -> 3876	2976 <- [3227 - 3321 - 3421] -> 3604	3608 <- [3623 - 3709 - 3793] -> 3825	4453 <- [4453 - 4523 - 4661] -> 4748	2559 <- [2655 - 2656 - 2660] -> 2674	3982 <- [4079 - 4213 - 4356] -> 4743	12183 <- [18509 - 19553 - 20811] -> 21515	15319 <- [15660 - 15796 - 15912] -> 16056	1517 <- [1920 - 1997 - 2103] -> 2113

❗ Test metadata have changed

--- /dev/fd/63	2021-11-04 11:58:41.981693399 +0000
+++ /dev/fd/62	2021-11-04 11:58:41.981693399 +0000
@@ -8,33 +8,33 @@
   },
   "systemInformation": {
     "cpu": {
-      "brand": "Xeon® E5-2673 v4",
+      "brand": "Xeon® E5-2673 v3",
       "cache": {
         "l1d": 65536,
         "l1i": 65536,
         "l2": 524288,
-        "l3": 52428800
+        "l3": 31457280
       },
       "cores": 2,
       "family": "6",
-      "flags": "fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology cpuid pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch invpcid_single pti fsgsbase bmi1 hle avx2 smep bmi2 erms invpcid rtm rdseed adx smap xsaveopt md_clear",
+      "flags": "fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology cpuid pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm abm invpcid_single pti fsgsbase bmi1 avx2 smep bmi2 erms invpcid xsaveopt md_clear",
       "governor": "",
       "manufacturer": "Intel®",
-      "model": "79",
+      "model": "63",
       "physicalCores": 2,
       "processors": 1,
       "revision": "",
       "socket": "",
-      "speed": 2.3,
+      "speed": 2.4,
       "speedMax": null,
       "speedMin": null,
-      "stepping": "1",
+      "stepping": "2",
       "vendor": "GenuineIntel",
       "virtualization": false,
       "voltage": ""
     },
     "mem": {
-      "total": 7289622528
+      "total": 7291699200
     },
     "osInfo": {
       "arch": "x64",
@@ -42,11 +42,11 @@
       "codename": "Focal Fossa",
       "codepage": "UTF-8",
       "distro": "Ubuntu",
-      "kernel": "5.11.0-1020-azure",
+      "kernel": "5.8.0-1040-azure",
       "logofile": "ubuntu",
       "platform": "linux",
       "release": "20.04.3 LTS",
-      "serial": "92bef7b423eb4bb0b05dd0208eb3d3da",
+      "serial": "cfc067bfcb844f35865e279a1b0e66c5",
       "servicepack": "",
       "uefi": false
     }

[trungleduc]

Thank @davidbrochart for your work on this, this is looking very nice!

[trungleduc]

One question about testing, can we make all tests run on both jupyter-server and fps ?

[trungleduc]

I think we might need docstring and typing for this function because using self as a function parameter is a little bit confusing.

[davidbrochart]

One question about testing, can we make all tests run on both jupyter-server and fps ?

I didn't look at tests, I think they are very tightly coupled with Tornado and jupyter-server.

[davidbrochart]

UI tests are now passing.

[trungleduc]

I think you need to update the screenshots for voila project of playwright

[trungleduc]

@davidbrochart should we close this PR? It looks like FPS won't be maintained anymore.

[davidbrochart]

I didn't remember we started that. Yes I'll close it and we will need to open a new PR when jupyter-server/jupyverse#277 is merged.