Warp10 tutorial - 2025/11/10

Warp10 uses tokens to grant read and write access (with more fine grained control possible). A token is generated from a .mc2 file in the tokens directory, that defines the application name (equivalent to a database - different applications do not see each other's data), expiry period, etc.

Copy demo-tokengen.mc2 to polytech.mc2 and edit this last file to change the application name from demo.CHANGEME to tp.polytech

Run ./bin/warp10.sh tokengen tokens/polytech.mc2 to generate access tokens. They will be printed as JSON on stdout, with 1 token for reading and 1 for writing. These tokens should be specified for the different queries/updates that we can do on the server, and are associated with the given application name.

[Question 1] Copy in the report the generated JSON (actually it is more so that you keep it as a reference)

Note that Tokens are valid for a fixed period of time (configurable in the .mc2 file).

All interaction can be done through the HTTP API (see available endpoints), but it requires to URL-encode the parameters, which is not convenient (but can be useful to automate things). In this tutorial, we will use the WarpStudio web application. Access it through https://studio.senx.io/ (note: this is the hosted version, you could also run the studio interface locally).

In the top-left selector, make sure that you are referencing the http://localhost:8080/api/v0 instance (and not the sandbox one). As described in the 00-Welcome tutorial, configure the interface to have the tokens available in every request:

Go to settings (⚙ gear icon) / Warp 10 instances / Click "Edit" on the localhost endpoint and insert the code that stores your tokens as RT (Read Token) and WT (Write Token) in the WarpScript Bootstrap field. This code will be executed before each query to the given server, and give a convenient access to the token values.

'XXX..XX' 'RT' STORE

'YYY..YY' 'WT' STORE

Note These commands simply define the RT and WT variables so that their values - the tokens - are easily accessible in the queries that you will do, with the $RT syntax.

Have a look and experiment for 10 minutes with the 01-Basic file to get a feel of the WarpScript syntax. The book icon (top right) in the studio gives access to the WarpLib reference guide which gives detailed information about each function. There is also a CheatSheet for quick reference.

You can save (Save icon in top right toolbar) your current editor content into the browser storage - do not hesitate to do it frequently with different names for the different steps.

To get a first confirmation that you actually uploaded data, go to the Explore tab in the Studio interface, make sure that the localhost server is selected, enter the read token, and check if Explore gives an output.

After reading the 02-find-and-fetch-data tutorial, list the defined data series (classname+label).

[Question 5] Give the WarpScript query, and its output (hint: FINDSETS)

[Question 6] Give the WarpScript query to fetch the different timeseries for the class water.level between 2020-01-01T00:00:00 and 2023-01-01T00:00:00 and store it in a ts variable. Add also $ts as last command to push the GTS on the stack, so that you can visualize it.

Note: when your output is a GTS or a list of GTS, then the DataViz and Tabular View are available to display the data in different forms.

Since there is a huge number of datapoints, the DataViz component has trouble processing it properly. Notice however the global characteristics for both GTS, and have a look at the timestamps for some datapoints.

Warp10 provides the LTTB / TLTTB functions for downsampling using the Largest Triangle Three Buckets algorithm (trying to preserve the overall shape of the data). Try to apply the TLTTB algorithm with a number of samples of 100, then 1000, then 2000. Observe the differences in perception of the nature of the data.

You can also explore the different types of chart types that are available (see selector on the right) and also the visualisation of the map along the chart - not very useful here, except to check that your geolocation conversion is OK.

[Question 7] Determine the size (number of points) of each GTS in number of elements as a list (hint: LMAP - you will have to use a macro and pay attention to parameters). Give the result and the query.

The number of samples is different between both GTS. To be able to compare them more easily, we will bucketize the data (to align on the same intervals) and fill the missing indexes between both series.

Bucketize the data using 1h interval, using the bucketizer.mean function. Notice how the timestamps of the datapoints are now aligned on hour values.

[Question 8] Give the query for bucketizing the list of GTS.

The timestamps are now aligned, but there are many missing values. To align both series, we may drop the non-matching points (see this blog post), or fill them with interpolated values. We will try to interpolate using the filler.spline interpolator.

Note: the FILL command expects 2 GTS on the stack (and not a list of GTS). To convert from the list of GTS to its components, use LIST-> DROP (because LIST-> outputs the list contents and the number of elements). Then you can apply the FILL command, and convert the result back to a list using 2 ->LIST.

[Question 9] Give the query for filling the missing values using the filler.spline interpolator. Once filling is done, give the number of samples in each GTS.