Pierre-Yves David <pierre-yves.david@octobus.net> [Thu, 02 Feb 2023 17:18:24 +0100] rev 50550
safehasattr: pass attribute name as string instead of bytes
This is a step toward replacing `util.safehasattr` usage with plain `hasattr`.
The builtin function behave poorly in Python2 but this was fixed in Python3.
These change are done one by one as they tend to have a small odd to trigger
puzzling breackage.
Anton Shestakov <av6@dwimlabs.net> [Wed, 31 May 2023 12:02:56 -0300] rev 50549
debug: `isinstance(a, x) or isinstance(a, y)` is `isinstance(a, (x, y))`
Anton Shestakov <av6@dwimlabs.net> [Wed, 31 May 2023 12:01:25 -0300] rev 50548
debug: update usage strings of debugignore and debugnodemap
Multiple files can be specified for debugignore.
debugnodemap does not take a revision argument.
Anton Shestakov <av6@dwimlabs.net> [Wed, 31 May 2023 12:00:21 -0300] rev 50547
debug: slightly improve wording on the InputErrors from the previous patch
Anton Shestakov <av6@dwimlabs.net> [Wed, 31 May 2023 11:30:33 -0300] rev 50546
debug: use InputError instead of CommandError for validating arguments
Pierre-Yves David <pierre-yves.david@octobus.net> [Fri, 26 May 2023 17:41:25 +0200] rev 50545
clonebundle: add a `filter_bundle_url` function
This function does nothing by default, but give extension the opportunity to
alter the URL, typically, this could be used to inject authentication token when
serving clone bundle for private repositories.
Pierre-Yves David <pierre-yves.david@octobus.net> [Fri, 26 May 2023 16:55:52 +0200] rev 50544
clonebundles: move the manifest reading in a dedicated function
We are about to make the logic more advanced to help hosting solution, so we
need to centralize it first.
Joerg Sonnenberger <joerg@bec.de> [Wed, 31 May 2023 22:36:50 +0200] rev 50543
doc: format argument for date uses strftime format string (
issue6818)
Mathias De Mare <mathias.de_mare@nokia.com> [Thu, 20 Apr 2023 11:23:45 +0200] rev 50542
clonebundles: filter out invalid schemes instead of failing on them
Previously, an invalid clonebundle scheme would result in a failed
clone. By specifying a list of schemes we support,
we can make sure adding a new scheme (like the one for inline clonebundles)
does not result in clones failing for older clients.
Mathias De Mare <mathias.de_mare@nokia.com> [Thu, 20 Apr 2023 10:48:12 +0200] rev 50541
clonebundles: demonstrate bad behaviour when unknown scheme is present