One thing came to mind while reading Frans comment: The driver should be actually be able to detect a language variant and report that information to the storage during indexing, e.g for the local driver this could mean that de.foo.txt
is the german variant of foo.txt
.
Philipp the issue is that the driver doesnât know the language. The same as it doesnât know anything of the other metadata we keep in sys_file_metadata. It basically only knows the values we store in sys_file.
Relying on a file prefix doesnât really sound like a solid solution. What if the editor changes the filename etc.
I donât think we need to have a automagic solution here. Just give the editor an option to set the contents_language and the related file
Philipp the issue is that the driver doesnât know the language. The same as it doesnât know anything of the other metadata we keep in sys_file_metadata. It basically only knows the values we store in sys_file.
Well, the driver creates the information in sys_file. It and only it can provide any information about the file. I do not say that the driver needs to provide this information, but - depending on the storage - it might have that information at hand (e.g. knows about variants, etc).
I donât think we need to have a automagic solution here. Just give the editor an option to set the contents_language and the related file
That is to error prone. Editors will forget it half of the time and it requires manual intervention, therefore troubling any automated process.
This topic was automatically closed after 31 days. New replies are no longer allowed.
This is indeed a major issue. We ran into the same problem already with custom records as well, where those should have been shared across trees with different default languages => not possible.
I disagree having any magic involved here. Iff the driver has more info (eg from a cdn or cloudâish storage) it should of course provide this metadata, but it must never make any conclusion. (unless you write a driver that is specifically tailored to a use case where a language can be detected due to a pattern in the filenames)
The big listing of @minifranske reads quite good, yet Iâm missing the hard implementation details, specifically when it comes to the relation with sys_file_reference
which is again translatable.
Please also write down the specifics you envision in this regard.
Thanks.