Auto-Saving implemented in r16227 (to become 2.1.5) - Feedback wanted

Christian Stimming stimming at
Tue Jul 3 05:49:53 EDT 2007

Quoting Beth Leonard <beth at>:
> Does it save over your current working datafile? Or does it save a
> backup snapshot of the datafile that the next time you open gnucash
> after a crash gnucash will ask you if you want to open the backup or
> the real thing?

In that sense: Yes, it does indeed "save over the current working  
datafile", say, foo.xac. But the old datafile will be kept around as a  
time-stamped backup copy, say, foo.20070702121341.xac.

This fact is admittedly not at all obvious. Nevertheless the older  
data does exist. However, from the time-stamped filenames you can't  
tell which one was saved by pressing "save" and which one by the  
current auto-save implementation.

> If gnucash had "undo" I could see having an autosave that is on by
> default as a good feature, but without undo, I'd much rather see it
> save a copy (like all those tmp files gnucash makes that don't
> include business features) instead of overwriting the current working
> file.  Right now the only chance a user has to undo things is to
> go back to the previously saved copy.

Hm... the problem here is that this would require major changes in how  
the file saving works, and also the program has to keep track not only  
of the "book-dirty state" but additionally of the "book-manually-saved  
state" (to distinguish whether the book has been auto-saved, or  
manually saved, or both). From a programmer's point of view this is  

The main reason for me to implement this so quickly was that I  
discovered in the current way the implementation was surprisingly easy  
from a programmer's point of view. What I'm saying is that if the  
feature in the current form doesn't meet user's needs, we can very  
well disable this again... sorry for that.


More information about the gnucash-devel mailing list