Merge lp:~wgrant/launchpad/reject-not-accepted into lp:launchpad
| Status: | Merged |
|---|---|
| Merged at revision: | 17882 |
| Proposed branch: | lp:~wgrant/launchpad/reject-not-accepted |
| Merge into: | lp:launchpad |
| Diff against target: |
60 lines (+3/-20) 2 files modified
lib/lp/archiveuploader/nascentupload.py (+2/-9) lib/lp/archiveuploader/tests/nascentupload.txt (+1/-11) |
| To merge this branch: | bzr merge lp:~wgrant/launchpad/reject-not-accepted |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Colin Watson | 2015-12-30 | Approve on 2016-01-08 | |
|
Review via email:
|
|||
Commit Message
Fix NascentUpload.
Description of the Change
Fix NascentUpload.
It would previously do so if the upload made it to Done before an exception occurred (eg. because build creation failed).
| William Grant (wgrant) wrote : | # |
On 05/01/16 01:57, Colin Watson wrote:
> Review: Needs Information
>
> This seems to change do_reject to only send the mail, and not
> actually change the status (PU.notify doesn't do that). That seems
> odd. Can you explain what's going on there?
The transaction is about to be aborted, so the only side-effect that can
depend on the status is the email. Since we can't reliably set the
status, I opted to be consistent and rely solely on the override.
| Colin Watson (cjwatson) wrote : | # |
OK, I hadn't traced things through to confirm that do_reject is always followed by an abort (and was put off by the original nascentupload.txt test that you removed; I now realise that it goes through a path that doesn't involve the abort). Now that I have, I agree with you.

This seems to change do_reject to only send the mail, and not actually change the status (PU.notify doesn't do that). That seems odd. Can you explain what's going on there?