Hey,
I am investigating the Translation Manager in Perspective. I am trying to find a way to synchronize projects with each other and the import functionality caught my attention.
When it comes to large files (70.000+) the import time seems to increase per new key.
I tested approaches like the Translation Manager import function, TranslationPackageDiff (with different chunk sizes) or the system.util.modifyTranslation function.
When I used the TranslationPackageDiff function, I was able to test the runtime and I discovered that the import time per 10.000 keys increases the more keys are stored in the Translation Manager. I assume that each key is checked for singularity(?).
Is there a more performant way to import language packages?
Here is some data to show what I mean 
| Time | Difference [s] | Key Count |
|---|---|---|
| 13:34:50 | 00:00:36 | 10000 |
| 13:35:26 | 00:00:47 | 20000 |
| 13:36:13 | 00:01:14 | 30000 |
| 13:37:27 | 00:01:35 | 40000 |
| 13:39:02 | 00:02:15 | 50000 |
| 13:41:17 | 00:02:40 | 60000 |
| 13:43:57 | 00:03:08 | 70000 |
| 13:47:05 | 00:03:40 | 80000 |
| 13:50:45 | 00:04:09 | 90000 |
| 13:54:54 | 00:04:39 | 100000 |
The import function uses the TranslationPackageDiff and every 10.000 keys I run the applyDiff function.
Kind regards
Janis