Re: Logging user actions

[Mikkel Kamstrup Erlandsen <mikkel kamstrup gmail com>]

On Tue, 2006-09-26 at 21:04 -0500, Federico Mena Quintero wrote:
> On Mon, 2006-09-25 at 11:23 +0200, Alexander Larsson wrote:
> > Sounds interesting. Maybe we could somehow get bug buddy to trigger a
> > log when nautilus crashes too?
> 
> Attached is my latest version of the patch; it's just polish over the
> last one.
> 
> This one also dumps the log when Nautilus crashes.  I'm dubious about my
> signal handling in nautilus-main.c, so I'd appreciate a quick review of
> that part from someone with more hardcore Unix-fu :)  From a quick test,
> it works --- I sent a SIGSEGV to Nautilus with kill(1), and it dumped
> the log all right *and* invoked bug-buddy as intended.
> 
> > Personally I'm not totally sure logging highlevel operations are helpful
> > for many bugs, but having more information is never bad.
> 
> What I want is to have some idea of how to reproduce things.  Over time
> I want to add more fine-grained logging (as well as logging for the
> other user actions, such as "copy a bunch of files", "edit the
> properties of a file", etc.).
> 
> For example, right now I don't have a way to distinguish "file was
> activated with Enter" vs. "file was activated with right-click/Open",
> etc.  In theory you shouldn't need that amount of detail.  In practice,
> we'll see --- it may turn out that we need to discriminate between those
> things.

I admit that I haven't really looked at the code, but I just had a
thought. Can this logging framework be used to implement an "Undo"
feature[1]?

Cheers,
Mikkel

[1]: With the obvious limitations of reverting permanently deleted files
and such things.