Feat: Slack bot integration #427

nulinspiratie · 2016-12-27T05:02:22Z

This pull request adds a QCoDeS slack bot that provides information about current measurements.
The user can communicate with the Slack bot via instant messaging (see image below).
The slack bot currently has the following functionality:

Give information about the current measurement (percentage done, print of dataset).
Upload the most recent plot
Notify when measurement is done
Adding custom commands and periodically-executed tasks

A few notes

I added the class attributes DataSet.latest_dataset and BasePlot.latest_plot. Every time a dataset/plot is instantiated, this attribute points to that object. This allows the slack bot to determine the latest dataset/plot.
I added to slack bot to qcodes/widgets, but I was not sure if this is the right place. If not, where should it go?
Periodic calls to Slack.update() must be made, and so the Slack bot works best as a widget in multiprocessing.
The package 'tempfile' is used to determine a temporary store location for any plot file. This isn't currently in the package requirements

@giulioungaretti @AdriaanRol

giulioungaretti · 2016-12-27T09:19:56Z

lol this is super cool! May need more massaging though, as dataset.sync() may disappear soon, and same goes for the hiddenupdatewidget !

We can definitely work a way to insert the slack class into the code base, I have the feeling that this should rather be a daemon-like process (such as monitoring) than a widget like thing :D

Terrific contribution !

nulinspiratie · 2017-01-03T04:19:52Z

@giulioungaretti I wasn't aware that dataset.sync() would disappear, will it not be necessary later on?

And I was initially also thinking of a subprocess, but with the current multiprocessing architecture I ran into problems with retrieving the latest plot. Also, I noticed some cool things you can do when the Slack bot is a widget instead of a subprocess. For instance, you can create a cell, fill it with a chunk of code, and then execute it. I'm using this to run measurements from Slack atm, although I'm not sure if this functionality should be part of the QCoDeS package

nulinspiratie · 2017-05-18T20:56:11Z

@giulioungaretti I realized this code doesn't need widgets nor multiprocessing. Instantiating it in a different thread is sufficient. Would this PR be useful for qcodes? If so I can make the changes. What folder would now be a good location for the code?

giulioungaretti · 2017-05-19T14:04:24Z

We can definitely add this! I would make an extra folder next to utils.

The new version of slack is a thread, and so does not need a ipython widget. Furthermore, the active_loop and active_data_set functions have been added, since slack needs access to the active loop and data_set. active_loop() will point to the active loop. this is set when calling loop.run(set_active=True), where set_active is True by default. In some cases a measurement can itself consist of measurements. For instance, a complex parameter could perform an entire measurement. To avoid that the active_loop is overwritten, the kwarg set_active=False can be added.

nulinspiratie · 2017-06-06T08:48:19Z

@giulioungaretti I upgraded the Slack class, it now is a subclass of threading.Thread. The main advantage here is that it operates as a separate thread without depending on ipython widgets.

Starting the Slack client is very easy, namely as such:

from qcodes.utils.slack import Slack
slack = Slack(**optional_commands)

optional_commands are commands that can also be executed aside from the standard ones. For instance, it could be a parameter, which we can then perform get/set operations on via Slack.

Since the Slack class needs access to the latest measurement and plot, I also added some functionality to BasePlot and loops.py. BasePlot now has the class attribute BasePlot.latest_plot, which is updated every time a new BasePlot object is created. This way, latest_plot always refers to the latest created plot.

For loops I did something similar: I created ActiveLoop.active_loop. Using this, I added the two functions active_loop() and active_data_set(), which calls active_loop(). When running a loop with kwarg set_active=True (True by default), it will set ActiveLoop.active_loop to itself for the duration of the measurement. This also works with Measure. In the cases where you have a Loop in which a complex parameter itself creates a Loop or Measure, you can avoid overwriting active_loop by adding set_active = False.

I noticed that active_loop and active_data_set are also useful in other ways. For instance, you can use the folder of the active dataset as a base folder, and any parameters that create Loops can create subfolders in the base dataset folder. I added the two functions to qcodes init, so they can be accessed easier.

I finally also added slacker as an optional package in the setup

giulioungaretti · 2017-06-06T10:13:54Z

looks good to me! @jenshnielsen @WilliamHPNielsen ?

jenshnielsen · 2017-06-06T11:08:42Z

Looks good to me

giulioungaretti · 2017-06-08T09:43:38Z

great will review the code and merge//ask for changes today !

giulioungaretti

@nulinspiratie small comments, the ready to merge!

giulioungaretti · 2017-06-13T10:18:53Z

qcodes/utils/slack.py

+from qcodes import active_loop, active_data_set
+
+
+def convert_command(text):


not sure why this? Isn't there one extra def () ?

Not sure what you mean, the try_convert_str is a function that's only called within convert_command, so I made it a subfunction.

@nulinspiratie shut, never mind I totally misread the code. Just catch the right Value error and then I can merge!

giulioungaretti · 2017-06-13T10:19:19Z

qcodes/utils/slack.py

+        try:
+            val = float(string)
+            return val
+        except:


just catch: ValueError here :D

giulioungaretti · 2017-06-13T10:19:23Z

qcodes/utils/slack.py

+        try:
+            val = int(string)
+            return val
+        except:


just catch: ValueError here :D

Author: Serwan Asaad <serwan.asaad@gmail.com> Feat: Slack bot integration (#427)

* Added slack bot * Added latest_dataset/latest_plot * refactor: moved slack.py to utils * feat: upgraded slack, added active_loop, active_data_set The new version of slack is a thread, and so does not need a ipython widget. Furthermore, the active_loop and active_data_set functions have been added, since slack needs access to the active loop and data_set. active_loop() will point to the active loop. this is set when calling loop.run(set_active=True), where set_active is True by default. In some cases a measurement can itself consist of measurements. For instance, a complex parameter could perform an entire measurement. To avoid that the active_loop is overwritten, the kwarg set_active=False can be added. * added slacker to extra packages * Removed .idea from PyCharm * fix: Added catching of ValueError

nulinspiratie added 2 commits December 27, 2016 15:44

Added slack bot

d842a17

Added latest_dataset/latest_plot

e0ad781

nulinspiratie changed the title ~~Slack bot integration~~ Feat: Slack bot integration Dec 27, 2016

giulioungaretti added the enhancement label Mar 9, 2017

nulinspiratie added 5 commits June 6, 2017 18:02

merge: with newest qcodes

af4431a

refactor: moved slack.py to utils

5c3bc24

added slacker to extra packages

5b7e6c4

Removed .idea from PyCharm

49b5c8b

giulioungaretti self-assigned this Jun 8, 2017

giulioungaretti approved these changes Jun 13, 2017

View reviewed changes

fix: Added catching of ValueError

18037bd

giulioungaretti merged commit 77f2d1c into microsoft:master Jun 15, 2017

giulioungaretti pushed a commit that referenced this pull request Jun 15, 2017

Generated gh-pages for commit 77f2d1c

cff15ef

Author: Serwan Asaad <serwan.asaad@gmail.com> Feat: Slack bot integration (#427)

nulinspiratie deleted the slack-bot branch June 21, 2017 02:02

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Feat: Slack bot integration #427

Feat: Slack bot integration #427

nulinspiratie commented Dec 27, 2016

giulioungaretti commented Dec 27, 2016

nulinspiratie commented Jan 3, 2017 •

edited

nulinspiratie commented May 18, 2017

giulioungaretti commented May 19, 2017

nulinspiratie commented Jun 6, 2017 •

edited

giulioungaretti commented Jun 6, 2017

jenshnielsen commented Jun 6, 2017

giulioungaretti commented Jun 8, 2017

giulioungaretti left a comment

giulioungaretti Jun 13, 2017

nulinspiratie Jun 15, 2017

giulioungaretti Jun 15, 2017

giulioungaretti Jun 13, 2017

giulioungaretti Jun 13, 2017

		from qcodes import active_loop, active_data_set


		def convert_command(text):

Feat: Slack bot integration #427

Feat: Slack bot integration #427

Conversation

nulinspiratie commented Dec 27, 2016

giulioungaretti commented Dec 27, 2016

nulinspiratie commented Jan 3, 2017 • edited

nulinspiratie commented May 18, 2017

giulioungaretti commented May 19, 2017

nulinspiratie commented Jun 6, 2017 • edited

giulioungaretti commented Jun 6, 2017

jenshnielsen commented Jun 6, 2017

giulioungaretti commented Jun 8, 2017

giulioungaretti left a comment

Choose a reason for hiding this comment

giulioungaretti Jun 13, 2017

Choose a reason for hiding this comment

nulinspiratie Jun 15, 2017

Choose a reason for hiding this comment

giulioungaretti Jun 15, 2017

Choose a reason for hiding this comment

giulioungaretti Jun 13, 2017

Choose a reason for hiding this comment

giulioungaretti Jun 13, 2017

Choose a reason for hiding this comment

nulinspiratie commented Jan 3, 2017 •

edited

nulinspiratie commented Jun 6, 2017 •

edited