failure to handle connection errors
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Bazaar |
Invalid
|
Medium
|
Unassigned | ||
PQM |
Triaged
|
High
|
Unassigned |
Bug Description
bzr is not completely clear, to API users, about whether 'NotBranchError' is the only error they need to catch.
I think this is a bug in bzr, because, http can do smart server messages, and a connection reset from one http client (because the server doesn't support POST), doesn't meant that svn's http handler won't work.
This is also a bzr-svn issue, 'how to signal *I couldn't tell*' about a branch.
We should decide how to manage this in bzr (e.g. 'always raise a subclass of NotBranchError, use different subclasses for cases like 'ConnectionReset' or 'raise ProbeFailed(
No handlers could be found for logger "bzr"
Permission denied (publickey).
Traceback (most recent call last):
File "/home/
manager.
...
File "/home/
branch_vcs = self.get_
File "/home/
if handler.
File "/home/
branch = Branch.
...
File "/home/
_read_more
"(and try -Dhpss if further diagnosis is required)")
bzrlib.
connectivity and permissions (and try -Dhpss if further diagnosis is
required)
Changed in bzr: | |
status: | New → Confirmed |
importance: | Undecided → High |
description: | updated |
description: | updated |
Changed in bzr: | |
importance: | High → Medium |
Changed in bzr: | |
status: | Incomplete → Invalid |
I don't really understand this bug:
> bzr is not completely clear, to API users, about whether 'NotBranchError' is the only error they need to catch.
I don't see why they would think that is the only error we could raise.