Age | Commit message (Collapse) | Author |
|
|
|
Don't suppress empty depends2: if there might be a preceeding non-empty
depends2: or requires:
This is neccessary to ensure setup's parsing code doesn't re-use a previous
non-empty value for the version under consideration.
|
|
|
|
|
|
|
|
always be replaced
This is intended for (temporarily) dealing with anomalous versions e.g.
those which have been removed (or relabelled as test), without a
superseding higher version existing, and indicates to setup that the named
versions(s), if installed, should be replaced with different version.
Also warn if this key pointlessly lists versions which will be replaced by
the current version anyhow.
|
|
|
|
|
|
|
|
|
|
depends: lines are recognized, but not handled correctly by setup versions
2.880 to 2.884 (they are ignored by setup version 2.879 and prior).
|
|
... and vice versa
Also, fix check that depends: packages exist to handle empty depends: (and
similarly for obsoletes:)
|
|
|
|
Add a disable-check: key in pvr.hint to disable certain checks, to gradually
supersede the static list in past_mistakes (which needs calm restarting to
re-read)
disable-check: contains a list of tokens, which indicates which checks are
to be downgraded from errors to warnings
in pvr.hint, disable-check: can contain:
empty-obsolete
check that if a binary package is empty and has no source package, it
is marked obsolete
unique-version
check that this binary package has the same version as all other
packages made from the same source package
These disable-check:s end up being applied to new versions when something is
split out from an existing source. That kind of suggests something is not
quite right about the test we are doing here. But we are limited because we
don't have a source-centric view of things due to source packages not really
being a thing yet...
|
|
|
|
Update tests appropriately
|
|
|
|
|
|
... as well as being repeated
|
|
Add a disable-check: key to hints to disable certain checks, rather than a
static lists in past_mistakes (which needs calm restarting to re-read)
disable-check: contains a list of tokens, similar to those for the
--disable-check option, which indicates which checks are to be downgraded
from errors to warnings for this package
in override.hint, disable-check: can contain:
curr-most-recent
check that the curr: version is the most recent non-test version
Also, don't consider equal mtime to be newer
Also, allow curr-most-recent to be disabled by --disable-check
|
|
Add mksetupini option --disable-check, as a more general replacement for
--okmissing
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Similarly, pprint of defaultdict was also fixed to print the defaultdict
repr, rather than as a dict
Explicitly convert to a dict in test_scan_uploads to keep the same text
representation
|
|
Shrink ldesc slightly in one hint to deal with not quite compatible choice
of where to break the line
|
|
Due to pprint internals, pprint of an OrderedDict() which fits on a single
line isn't effected by the monkey-patch. Make the sdesc in the one hint
which is rendered this way a little longer.
|
|
pprint was fixed by [1] from printing an OrderedDict as a dict, with the
lines in order, to printing the repr of the OrderedDict.
Monkey-patch pprint on python 3.6 to get the python 3.4 behaviour. Probably
better to do that the other way around, but that means changing the test
data...
[1] https://bugs.python.org/issue23775
|
|
This is just for the benefit of automation, as this license is already
present in each source file...
|
|
|
|
|
|
|
|
Converting a dependency atom to a package name with full generality requires
a database of all the pathnames contained in all packages, which we don't
have (yet)
For the moment, pass through dependency atoms which are cygwin package
names, and filter out all other kinds of dependency atoms
Also check that build-depends: is being applied to a source package
|
|
The archive filenames appear in setup.ini unchanged, but the version is
taken from the hint, rather than deduced from the archive filename.
This requires a change to collecting tars per VR, rather than keeping them
all in one list.
This could be used as a way to fix the epoch of packages which have a
decreased version number, but were built without bumping the epoch.
This could also be used to have a properly ordered version number whilst
using upstream version numbers which don't do that for the archive
filenames.
|
|
It's complex and repeated in many places.
|
|
This rejects a variable named 'l', in case it gets confused with '1'
|
|
|
|
|
|
|
|
|
|
Don't want components to be sorted with any leading whitespace
|
|
Future work: We probably need something to set requires: hint based on
depends: hint, if one isn't present.
|