bzr doesn't gracefully handle interrupted push/branch operations
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
| Bazaar |
Medium
|
John A Meinel |
Bug Description
If a keyboard interrupt, network/system failure, or any other event interrupts a branch or push operation, bzr is unable to resume or even gracefully re-initiate the previous operation. Instead, cryptic errors are presented to the user (no WorkingTree for interrupted branch, File Exists for push). Users have to remove .bzr by hand for bzr to function properly again.
This is terribly confusing to the novice user and annoying to the experienced user. Bzr should either
(a) Gracefully recover from interrupted actions, either by resuming from where it left off (preferable) or even cleaning out its previous attempt and starting from scratch
or (b) Recognize this situation and ask the user in an understandable message to remove .bzr
David Allouche (ddaa) wrote : | #1 |
David Allouche (ddaa) wrote : | #2 |
I mean bug 53340.
David Allouche (ddaa) wrote : | #3 |
or of bug 30576?
John A Meinel (jameinel) wrote : | #4 |
The associated branch has a fix which creates the branch before populating it, which should allow resuming the transfer.
Changed in bzr: | |
assignee: | nobody → jameinel |
status: | Unconfirmed → In Progress |
status: | In Progress → Fix Committed |
Robert Collins (lifeless) wrote : Re: [Bug 45504] Re: bzr doesn't gracefully handle interrupted push/branch operations | #5 |
On Thu, 2006-12-21 at 22:34 +0000, John A Meinel wrote:
> The associated branch has a fix which creates the branch before
> populating it, which should allow resuming the transfer.
I'm not sure this is a 'fix'. Surely just changing 'push' to create a
branch object if needed is simpler, and wont have the effect of having
an empty branch appear during the initial push.
--
GPG key available at: <http://
Changed in bzr: | |
status: | Fix Committed → Fix Released |
Is that a duplicate of bugj 53340?