Merge lp:~mwilck/duplicity/0.7-series into lp:~duplicity-team/duplicity/0.7-series
Status: | Merged |
---|---|
Merged at revision: | 1232 |
Proposed branch: | lp:~mwilck/duplicity/0.7-series |
Merge into: | lp:~duplicity-team/duplicity/0.7-series |
Diff against target: |
111 lines (+31/-19) 5 files modified
duplicity/globmatch.py (+16/-12) duplicity/selection.py (+3/-6) testing/functional/__init__.py (+1/-1) testing/gnupg/README (+2/-0) testing/gnupg/gpg.conf (+9/-0) |
To merge this branch: | bzr merge lp:~mwilck/duplicity/0.7-series |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
duplicity-team | 2016-07-27 | Pending | |
Review via email:
|
Description of the change
This merge request contains the same change as
https:/
for the 0.7 series this time. I believe a factor-20 speedup could be counted as a "bug fix".
Moreover, it includes 2 fixes that were necessary on my system (OpenSUSE tumbleweed) to make
the test suite pass. See the commit logs for details. I needed this in order to verify that the
first change didn't introduce any regressions.
- 1232. By ken on 2016-07-28
-
* Merged in lp:~mwilck/duplicity/0.7-series
- Speedup of path_matches_glob() by about 8x. See
https://code.launchpad .net/~mwilck/ duplicity/ 0.7-series/ +merge/ 301332
for more details.
Kenneth Loafman (kenneth-loafman) wrote : | # |
Martin Wilck (mwilck) wrote : | # |
Hmm, that would mean that setsid behaves differently on different distributions. It might be a difference in how setsid behaves, or a difference in pexpect.
In which test did the problems with "setsid -w" occur on precise / trusty?
Kenneth Loafman (kenneth-loafman) wrote : | # |
All of the tests in tests/functional, as far as I can tell.
On Thu, Jul 28, 2016 at 1:59 PM, Martin Wilck <email address hidden> wrote:
> Hmm, that would mean that setsid behaves differently on different
> distributions. It might be a difference in how setsid behaves, or a
> difference in pexpect.
>
> In which test did the problems with "setsid -w" occur on precise / trusty?
>
> --
> https:/
> You are subscribed to branch lp:duplicity/0.7-series.
>
Martin Wilck (mwilck) wrote : | # |
To make sure I understand correctly - my changes work on Xenial and Wily but not on older Ubuntu releases? Are you using Ubuntu native pexpect packages or pexpect from PyPi?
Kenneth Loafman (kenneth-loafman) wrote : | # |
I use pip, Launchpad uses their repo, which is out of date.
On Thursday, July 28, 2016, Martin Wilck <email address hidden> wrote:
> To make sure I understand correctly - my changes work on Xenial and Wily
> but not on older Ubuntu releases? Are you using Ubuntu native pexpect
> packages or pexpect from PyPi?
>
>
> --
> https:/
> You are subscribed to branch lp:duplicity/0.7-series.
>
Martin Wilck (mwilck) wrote : | # |
That might be the reason, then? Maybe we need to put in a version check on the pexpect module?
Martin Wilck (mwilck) wrote : | # |
The problem is caused by different syntax of setsid invocation. setsid on Trusty doesn't support -w.
I am pushing a patch that tests for -w support and uses the flag only if supported.
Aaron Whitehouse (aaron-whitehouse) wrote : | # |
For reference, while this talks about a x20 speedup, for me there is no material speed difference for the test suite between r 1231 and r 1232, so this speedup must be in a scenario not exhibited in the tests.
I had to remove -w from setsid to get precise and trusty to build on Launchpad.