Code review comment for lp:~manishsinha/zeitgeist-datasources/bzr-improvements-subjects

Am 24/03/12 17:54, schrieb Manish Sinha (मनीष सिन्हा):
>> The other subjects shouldn't be set if there is no working tree (in that case the files don't exist on disk).
> Can you give some examples?
You can have a branch without a working tree. E.g.:

$ bzr branch --no-tree -r5 lp:bzr-stats

will create a branch without a working tree. Running "bzr pull" there
later will pull in revisions but won't actually touch any of the
subjects the plugin now adds (as they don't exist).

>
>> I think the other subjects should only be set in the case of push and pull, and not in the case of commit as commit doesn't actually change those files.
> The think the reverse. In a commit there are various files involved. A file can be added/deleted/modified which form a part of an commit (which is an event)
The file itself doesn't change in that case though, only the branch
does. In what way is it the subject?
> In case of push and pull, we get the remote URI like bzr+shh:// which is not a valid URI for Gio.File due to which mimetype cannot be fetched.
That's not a problem - the working tree of remote branches isn't updated
by push and pull, only that of local branches.

>> It would make sense to add a post_merge hook too
> It would be good, but right now my plan was to get these three events and try to write a small client trying to understand how the information can be used. I have already trimmed XChat dataprovider as it was logging too much and was spamming the database with useless events. Before going ahead and adding merge/revert hooks, it would nice to know how we can use those events.
I think the same applies to adding too much subjects though... ?

Cheers,

Jelmer