This implements fixes for a couple of bugs that I filed to
make it easier to use desktopcouch in testing other projects.
The first use-case is testing against a private desktopcouch
for projects that use it, without stomping on the information
stored in gnome-keyring.
After that I wanted to re-use desktopcouch's knowledge of stopping
and starting couchdb in a project where I wanted to use plain couch,
not a desktop one, so I added with_auth to just have it skip all
the credentials stuff and just start a couchdb in a private area
for testing.
There aren't any tests for it, for two reasons:
* The setup of a private context is done at import time, which
is then re-used. I could do a similar thing in a couple of tests
to change the arguments, but...
* The tests themselves are not isolated from gnome-keyring. Therefore
I wondered if they wanted to make use of the with_keyring argument,
though that would mean that codepath wasn't tested.
Therefore I would like some guidance on what the preferred testing
strategy would be for this.
Hi,
This implements fixes for a couple of bugs that I filed to
make it easier to use desktopcouch in testing other projects.
The first use-case is testing against a private desktopcouch
for projects that use it, without stomping on the information
stored in gnome-keyring.
After that I wanted to re-use desktopcouch's knowledge of stopping
and starting couchdb in a project where I wanted to use plain couch,
not a desktop one, so I added with_auth to just have it skip all
the credentials stuff and just start a couchdb in a private area
for testing.
There aren't any tests for it, for two reasons:
* The setup of a private context is done at import time, which
is then re-used. I could do a similar thing in a couple of tests
to change the arguments, but...
* The tests themselves are not isolated from gnome-keyring. Therefore
I wondered if they wanted to make use of the with_keyring argument,
though that would mean that codepath wasn't tested.
Therefore I would like some guidance on what the preferred testing
strategy would be for this.
Thanks,
James