In large GitHub repositories the synchronisation becomes much slower.
This seems like Crowdin’s GitHub Integration just git clones the entire repository.
As most devs know if they’ve read up on Git, git clones are able to minimize the download size by adding the --depth 1 flag.
I’ll tried asking support but they just accept it as a fact, not an issue.
IMO the current integration adds a waste of resources (server, time, bandwidth) that negatively impacts both Crowdin and their Projects.
Possible counter argument
The argument could be made that Crowdin needs to compare the current state with that of the previous state: but even then Git allows pulling individual commits.
Putting developer time into this is a waste for such a small edge case: Not only would this lower server costs, the users will appreciate a much faster response from git pushes, improving unseen quality of life metrics.
As you said that Crowdin’s “GitHub integration” uses their api (not git) to retrieve the newest translation file, I tried it out myself & personally did not have the same slowdowns as that of Crowdin;
The only slowdown I perceived was that of getting to a file multiple directories down, but that would not be dependent on the repositories’ total size.
To get the branch’s (e.g. main) latest commit: docs.github. com/en/rest/git/refs?apiVersion=2022-11-28#get-a-reference
From there you get the root tree from that commit: docs.github. com/en/rest/git/commits?apiVersion=2022-11-28#get-a-commit-object