#7713 NORM 8.2.0 (: Inconsitent behavior for activity installation
Zarro Boogs per Child
bugtracker at laptop.org
Wed Aug 6 14:21:16 EDT 2008
#7713: Inconsitent behavior for activity installation
----------------------+-----------------------------------------------------
Reporter: erikos | Owner: marco
Type: defect | Status: new
Priority: normal | Milestone: 8.2.0 (was Update.2)
Component: sugar | Version: Git as of bug date
Resolution: | Keywords: 8.2.0:?
Next_action: design | Verified: 0
Blockedby: | Blocking:
----------------------+-----------------------------------------------------
Comment(by Eben):
Replying to [comment:4 cscott]:
> Having looked at the code, it doesn't seem like we actually support
having multiple versions of the same activity installed yet. We would
need to tweak the directory into which the .xo unpacks as well as add
'version' parameters to a number of API methods which currently only take
bundle_ids.
That's right.
> So it seems like 'upgrade' is the right solution for now. Hopefully in
9.1 we can fix the API issues and actually support multiple versions
concurrently.
Definitely. We've recently been discussing that, upon activity upgrade,
the bundle should also be starred according to the previously installed
version.
> But I'm not entirely clear what question erikos is asking. There are
some security issues involving bundle installation that require an
affirmative action by the user in the journal, but michael assures me that
installing an activity does involve executing any of its code, and so it
should be safe just to install/upgrade the activity as soon as it is
placed in the Journal. (In the 9.1 time frame I hope that we will be able
to execute activities directly from their representation in the datastore,
avoiding this 'installation' step.)
This is not true now? The whole point of exposing the bundle directly in
the Journal is to allow one to open a "clean instance" of the activity.
What happens when you click on a bundle now, if it doesn't launch the
activity?
> BUT for library bundles, installing does involve executing code, so it's
not necessarily safe to install them without explicit confirmation from
the user. (This is a bug; installing libraries should not require
executing code.)
Yeah, good point. I wasn't aware of this.
--
Ticket URL: <http://dev.laptop.org/ticket/7713#comment:5>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system
More information about the Bugs
mailing list