Merge lp:~kissiel/checkbox/fix-1501354-back-after-testplan-selected into lp:checkbox
Proposed by
Maciej Kisielewski
Status: | Merged |
---|---|
Approved by: | Zygmunt Krynicki |
Approved revision: | 4043 |
Merged at revision: | 4041 |
Proposed branch: | lp:~kissiel/checkbox/fix-1501354-back-after-testplan-selected |
Merge into: | lp:checkbox |
Diff against target: |
95 lines (+23/-11) 3 files modified
checkbox-touch/components/CheckboxTouchApplication.qml (+5/-1) checkbox-touch/py/checkbox_touch.py (+13/-0) plainbox/plainbox/impl/session/assistant.py (+5/-10) |
To merge this branch: | bzr merge lp:~kissiel/checkbox/fix-1501354-back-after-testplan-selected |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Zygmunt Krynicki (community) | Approve | ||
Nicholas Skaggs | Pending | ||
Review via email: mp+273460@code.launchpad.net |
Description of the change
This MR fixes a problem, where navigating back to select different test plan would break the app.
aa66fbf plainbox:
112da1b checkbox-touch: let changing of test plans
9ba76c1 checkbox-touch: update sessionDir after selecting new test plan
3f6b13b plainbox:
2ed4471 plainbox:
To post a comment you must log in.
+1
For some context:
09:58 <@zyga> kissiel: https:/ /code.launchpad .net/~zyga/ checkbox/ sa-api- changes/ +merge/ 273502 /code.launchpad .net/~kissiel/ checkbox/ fix-1501354- back-after- testplan- selected/ +merge/ 273460 test_plans (read only) test_plan: ... test_plans is exactly that
09:58 <@zyga> kissiel: have a look and tell me what you think
09:58 < kissiel> ack
09:58 <@zyga> kissiel: I'll propagate this down to all the apps if you ack
10:00 < kissiel> zyga: those pep8 got fixed on the branch I proposed yesterday
10:01 <@zyga> kissiel: heh
10:01 <@zyga> kissiel: where? let's merge em
10:01 < kissiel> zyga: https:/
10:02 * zyga reads
10:02 <@zyga> kissiel: can you ask the sa if a test plan was set?
10:02 <@zyga> kissiel: I'd like to minimize the state that apps need to have
10:03 < kissiel> zyga: I tried that approach
10:03 < kissiel> zyga: and do all the logic in SA
10:03 <@zyga> kissiel: what is the timestamp for?
10:03 < kissiel> zyga: there are problems with resuming session
10:03 <@zyga> kissiel: I remember that
10:04 <@zyga> kissiel: what I mean specifically is why don't we expose sa._manager.
10:04 < kissiel> zyga: timestamp goes to the app-blob
10:04 <@zyga> kissiel: so that app doesn't need to keep this in this case
10:05 < kissiel> zyga: the thing is you can have boostrap already done
10:05 <@zyga> kissiel: we already track test plan selection
10:05 < kissiel> zyga: and we want to get rid of all the progress
10:05 <@zyga> kissiel: how does bootstrap prevents this?
10:05 < kissiel> zyga: and I wouldn't like to expose job-state-map
10:05 <@zyga> kissiel: I understand, you want a new session
10:05 <@zyga> kissiel: all I'm saying that you store something wa already store
10:06 < kissiel> zyga: ok, so once again, app stores it, because when resuming session app will have it set to none
10:06 <@zyga> kissiel: c-c stores test plan that sa already keeps track of there, that if could have been: if sa.selected_
10:06 < kissiel> zyga: app differentiates between resuming and starting
10:06 < kissiel> sa does not
10:06 <@zyga> kissiel: are you sure?
10:06 < kissiel> zyga: spent quite a bit of time on that yesterday
10:06 <@zyga> kissiel: it pretty much seeems that sa._manager.
10:07 <@zyga> kissiel: what am I missing then?
10:07 < kissiel> zyga: when the session is resumed, sa will start a session out of a checkpoint, right?
10:08 < kissiel> zyga: and when we change the testplan (by navigating back and forth) bootstrapping is done for each tp selected
10:08 < kissiel> and once you go back, you need to restat
10:08 < kissiel> when resuming, we couldn't see if we need to drop the progress already done or not
10:08 < kissiel> becase we 'just already have a test plan selected in sa'
10:11 <@zyga> kissiel: I need to look into this
10:11 <@zyga> kissiel: I think we should not need to re-boostrap the whole way in this case, it's just unfortunate effect of how the gui design and API doesn't let you do what you want
10:12 <@zyga> kissiel: I think that what we have now is coarse-grained for the problem (we nuke and restart which is expensive)
10:12 < kissiel> zyga: yeah, it would be nice if session could stay, and tp could change
10:13 <@zyga> kissiel: I think we ...