The ApiClientΒΆ

The ApiClient communicates with the Coscine REST API and enables the user to make requests to and get responses from Coscine.


To speed things up it makes use of a request cache that stores responses from Coscine for re-use. As a consequence not every function provided by the Coscine Python SDK that makes a request to Coscine actually sends a request. Some may only send a request once and later on revert to the cached response. Other users in Coscine may make changes in the meantime, thereby effectively invalidating cached data. To make sure that this does not lead to inconsistencies, only constant data is cacheable and even that kind of data has a limited lifetime in the cache. Once that lifetime ends, the request is automatically refreshed on the next access.
Since constant data in Coscine rarely changes, the cache may be saved to file and re-used in later sessions. This can be disabled by the user.

Initialization and configuration 🎒¢

All we need to get started after we have installed python and the coscine package is these few lines of code:

import coscine

token = "Our API Token"
client = coscine.ApiClient(token)

Parallelization πŸš€ΒΆ

While it does not come with builtin support for parallel requests, the Coscine Python SDK can easily be used with concurrency in mind using external measures such as a thread pool. The following snippet uses the ThreadPoolExecutor type provided by the standard library module concurrent. In that snippet the executor manages 8 threads to which it delegates function calls. We no longer have to wait until one function returnes but instead can have multiple functions run at the same time - in this case 8, since there are 8 threads. By increasing the number of threads we can increase the number of operations that we are able to do concurrently. However, computing resources are generally scarce and we also should not send too many requests at once to Coscine in order to not trigger the rate limiter and get a temporary timeout. A reasonable measure for the amount of threads could be a low multiple of the amount of processor cores that your current machine has to offer or even the exact amount of cores. Anything in the order of 2 to 16 threads will suffice.

from concurrent.futures import Future, ThreadPoolExecutor, wait

emails = ["", "", ...]
with ThreadPoolExecutor(8) as executor: # 8 threads
    futures = []
    for email in emails:
        future = executor.submit(project.invite, email)
# The tasks are not running in several threads. We are still in the main
# thread at this point and can either process something else in the
# meantime or wait for the threads with our function to finish.
# Once they are finished we can process the results.
# All futures arrived at this point since we were waiting for them
for future in futures:
    function_return_value = future.result()

If you need the results returned by these functions you need to wait for their futures. Futures are not returned immediately since the function is running in another thread. Therefore the main thread does not get the result of the function in time and proceeds executing whatever comes next. The result therefore arrives in the future. When we are done sending stuff to Coscine and want to process the results, we can wait for the remaining futures to appear and then process them.

For more information on launching parallel tasks with python refer to the official documentation.


The Client uses caching internally. Sometimes you may run into cache consistency issues.
To clear the cache use: