[digiKam-users] Read Error in windows version but not in linux version

[Gilles Caulier]

2018-07-09 2:57 GMT+02:00 Simon Cropper <scropper at botanicusaustralia.com.au>
:

>
> In my mind this seems to imply that your directory, file names and maybe
> even your associated metadata have characters not shared between Linux and
> Windows.
>
> Linux uses UTF-8 while Windows CP-1252. They are different code pages and
> when characters unique to one or other code page is used in a system that
> is not code-page aware -- like Windows, the programs get upset.
> The only way to resolve the issue is to fix the offending paths, file
> names or metadata.
>
> This is applicable for any program and is not unique to DigiKam.
>
> Although this may not be the main problem, my understanding is that the
> path of files are stored as UUID+Path. I am not sure how robust the
> conversion routines are converting paths between Linux/UNIX format and
> Windows Format. Some of the issues may relate to this. Certainly most
> people report problems when moving their data between drives as this
> changes the UUID.
>
> If I have the time, to test, I'd create one directory with a small number
> of files. Check all of them for any character that is are not a-z, A-Z,
> 0-9, space and _
>
> Duplicate on your NAS.
>
> Try creating a database in Linux using one of the directories.
> Try creating a database in Windows using the other directory.
>
> Try adding the Linux directory to Windows install and visa versa. I
> suspect this won't work. If it does, then the issue you are reporting is
> probably just code page incompatibility and you need to seek out the
> offending characters.
>
> *@Gilles -- This list regularly gets queries about how to setup a central
> photo repositories and have multiple clients looking into the a single
> database. In most cases, like Clive its involves mixed operating systems. I
> have never seen a clear response from the developers if this is even
> feasible. Can you please clarify?*
>
>
Sharing the database between different Operating system is problematic.
Characters encoding is one, event if this can be fixed to use UTF8
everywhere. Qt is able to converter from one encoding to another one easily.

But the most important is enable database lock mechanisms to prevent
concurrent access at the same time from different computers. It's can be
complex and introduce limitation certainly in this kind of use. We have
already few entries in bugzilla about this topic.

In others words, i don't recommend to play in this configuration for the
moment, or at least in only this case :

- same type of OS access to a centralized the database (Mysql)
- only one computer use the database, no more than one at the same time.
- collection are shared on the network, and UUID to identify the collection
are the same everywhere (else data will be duplicated in database).

Best

Gilles Caulier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/digikam-users/attachments/20180709/52e75690/attachment.html>