I had some trouble after doing a reinstall of Linux, with all programs on one partition and
the home directory on another (where also CT is installed), including
Oracle Java. Similar things might happen on other OSs (Windows), I
suppose. And it might also happen when upgrading a Linux installation.
What have I done then?
So Linux users have their data spread
For Mac users, it is even a bit trickier to use "show package content" (a thing that most beginners won't do), but at least CT can be set into the Programs folder.
Reinstall of the whole program package (instead of the updates) in any OS might and/or will delete a certain amount of your individual data.
Wouldn't it be a great idea to have finally all data in one subfolder (mostly a hidden folder .cafetran in the user directory)?
I also wonder whether the Total Recall data - of let's say 1 or 2 GB - is being loaded into the RAM, even if it is not used, when it is part of the program package on the Mac.
Torsten: For Mac users, it is even a bit trickier to use "show package content" (a thing that most beginners won't do), but at least CT can be set into the Programs folder
I wonder if what you want cannot be achieved by adding the CT app/package to each client/project. I think that's possible on a Mac, not possible under Windows, Linux me not know at all.
Total Recall data - of let's say 1 or 2 GB - is being loaded into the RAM
It is not. That is, not the database. A resulting TMX file is, of course.
Torsten: Not really an option. 100 projects = 100 CT installations?
I meant recurring projects, of course. My suggestion could work for a few recurring projects.
CT without the database is still very small - less than 20 MB? - and the database does not have to be part of the package.
Increasing CafeTran's portability between the OSes is a good idea. It needs to be approached with special care though as the risk of loosing access to a resource is quite high during the OS switch. You can already access various resources via the could (e.g. Dropbox) from different systems. Even, Total Recall database can be configured in Preferences to access it from outside CT default installation folder. The newly added templates feature can also be used across the operating systems if you set the templates folder in Dropbox. There is definitely some room for improvement in this area.
I had to perform a clean install the GNU/Linux recently, and CT worked fine after restoring data from Backintime (like TIme Machine). CT is inside a specific folder in my home folder, along with some other apps. You can create dekstop file and have a launcher appear in the menu, etc.
What kind of settings are stored in the ~/.java folder for CT?
OmegaT has a nice implementation I think for preferences (stored in ~/.omegat folder for Linux and see screenshot for locations depending on the OS version), as a Java app it may be worth a look.
> reinstalling the OS (in this case Linux) without any change of locations or settings leads to a non-functional installation (this was the trigger for this posting); this could only be arranged by deleting/renaming the old .java folder (and then, why are CT settings, e.g. the list of last projects, language pairs etc. in this particular subfolder?)
That might have been caused by Linux imposing its own (open source) version of Java. I hope that in coming Java 9, Oracle's Java and Open JRE/JDK will be more compatible.
Indeed omegaT (having a integrated Java and putting everything into the subfolder .omegat) might lead the way.
I did the same again (reinstall OS the Linux way, with an own /home partition, same JAva and so on. Now it worked lke a charm.
Nevertheless, a all-in-one user directory (i.e. ".cafetran" for all settings) and a more flexible way to export and import settings or part of the settings would be nice.