Without the non-translatables glossary, auto-propagation works as advertised:
Can you confirm this abnormality?
Can you fix it?
Yes, I can confirm that non-translatable fragments based on regular-expressions can't be auto-propagated. Regular expressions are designed to catch phrases that are not necessarily identical. They can be similar (e.g a group of numbers) and auto-propagation function can't be applied to the matched phrases that may be similar in various segments. It needs to operate on principle of identity.
For example, if one segment contains a matched number based on the regular expression \d+, the same regular expression can catch quite a different number in another segment so it can't be auto-propagated from the first segment. Just try to use non-translatable fragments which are not regular expressions.
There's no easy way for you to add number adaption to enable propagation, is there?
So, if I have 1000 error messages, starting each with F + a three-digit number, I should add F001 ... F999 "hard coded" to my non-translatables glossary, in order to make auto-propagation possible?
I will test this!
You can create such a list easily as follows:
1. In the Find panel, select Regular expression and Extract reg. exp. results.
2. Test your regular expression in the Find field and click the Find button.
3. CafeTran will create the list for you.
I would have used Ms Excel, but your solution is easier.
alwayslockyourbike
If a glossary with these settings is assigned to a project:

and auto-propagation is set to:
while your non-translatables glossary contains exactly one entry:
Auto-propagation won't take place:Can this be fixed, please?