- 30 8月, 2020 2 次提交
- 19 4月, 2020 1 次提交
-
-
由 Ivo De Decker 提交于
-
- 18 4月, 2020 3 次提交
-
-
由 Ivo De Decker 提交于
-
由 Ivo De Decker 提交于
In python3, these are iterators, which is fine in most situations. This commit changes that places where it's not.
-
由 Ivo De Decker 提交于
-
- 05 1月, 2020 1 次提交
-
-
由 Niels Thykier 提交于
The code base contains a few unfornuate patterns such as: data = open(...).read() OR fh = open(...) data = fh.read() This causes a file descriptor leak as the opened file is never explicitly closed again. While slightly mitigated by CPython using ref-counting and immediate closure, we should still update the code to close the files. This commit focuses on a large number of simple clean ups. Most of which relies on the "with open(...) as X" pattern where feasible. As "with" causes identation noise, this commit deliberately refains from "fixing" other suboptimal behaviour. Notably there are several instances of "except:" visible in the diff that should ideally be reduced. Signed-off-by: NNiels Thykier <niels@thykier.net>
-
- 30 11月, 2019 1 次提交
-
-
由 Paul Wise 提交于
It is shorter and more readable.
-
- 20 10月, 2019 1 次提交
-
-
由 Ansgar 提交于
-
- 16 9月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 04 9月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 16 7月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
This was a regression introduced in 77dc9328Reported-by: NGianfranco Costamagna <locutusofborg@debian.org>
-
- 06 7月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 28 6月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 11 6月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 10 6月, 2019 2 次提交
-
-
由 Ansgar Burchardt 提交于
-
由 Ansgar Burchardt 提交于
The only difference between the two versions was the function used to parse the dependency.
-
- 04 4月, 2019 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 10 11月, 2018 2 次提交
-
-
由 Ansgar Burchardt 提交于
-
由 Guillem Jover 提交于
For source-only uploads the fields should have been empty all along, because there are no binary packages to include, but dpkg-genchanges has incorrectly included all binary names that can be produced from the debian/control template file (see #818618). Although DAK allows an empty Binary field, the correct behavior should be for the field to not be present if there are not binary packages to upload.
-
- 06 11月, 2018 1 次提交
-
-
由 Guillem Jover 提交于
The affected packages broken by that field value are fixed since Debian oldstable (Jessie). See also #759099. Closes: #796213
-
- 28 6月, 2018 2 次提交
-
-
由 Bastian Blank 提交于
-
由 Bastian Blank 提交于
-
- 27 6月, 2018 1 次提交
-
-
由 Bastian Blank 提交于
-
- 25 5月, 2018 1 次提交
-
-
由 Bastian Blank 提交于
-
- 24 5月, 2018 1 次提交
-
-
由 Bastian Blank 提交于
-
- 22 5月, 2018 2 次提交
-
-
由 Bastian Blank 提交于
-
由 Bastian Blank 提交于
-
- 21 5月, 2018 2 次提交
-
-
由 Bastian Blank 提交于
-
由 Bastian Blank 提交于
-
- 20 5月, 2018 2 次提交
-
-
由 Bastian Blank 提交于
-
由 Bastian Blank 提交于
-
- 19 5月, 2018 2 次提交
-
-
由 Bastian Blank 提交于
-
由 Chris Lamb 提交于
-
- 08 4月, 2018 1 次提交
-
-
由 Ansgar Burchardt 提交于
This will be used by the signing service.
-
- 08 3月, 2018 1 次提交
-
-
由 Julien Cristau 提交于
This lets us reject an upload up front rather than dying later in process-upload.
-
- 02 11月, 2017 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 15 7月, 2017 1 次提交
-
-
由 Ansgar Burchardt 提交于
-
- 06 7月, 2017 2 次提交
-
-
由 Ansgar Burchardt 提交于
Some other checks use it to see which architectures are included, for example the previous commit.
-
由 Ansgar Burchardt 提交于
For policy queues dak keeps the .buildinfo and .changes around until the upload is processed. As the uploads might already be sent to buildds while waiting in the policy queue, the buildd upload might use the same filename and end up getting rejected. With this change, the upload getting rejected should be the maintainer's upload.
-