Debian Patches
Status for tellico/4.1.4-2
| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| upstream_Adapt-XML-validation-tests-to-libxml2-2.15.patch | [PATCH] Adapt XML validation tests to libxml2 2.15 It looks like libxml2 2.15 enhances the validation check done by xmlValidateNCName() (used by the helpers in Tellico::XML), accepting also the character "€" (i.e. EURO SIGN). Considering that "ě" (i.e. LATIN SMALL LETTER E WITH CARON) was already accepted, this seems a valid behaviour change. Hence tweak the test a bit, providing slightly different test data for the newer libxml2 version, including the EURO sign as valid character. |
Pino Toscano <pino@kde.org> | no | 2025-10-27 | ||
| skip-failing-filelisting-test.diff | no | |||||
| tests-avoid-Group_Statistics.diff | Do not use Group_Statistics.xsl as test data It is a buggy template, and it was removed upstream; will not be there anymore in 4.1.5+. . See https://invent.kde.org/office/tellico/-/commit/6f765a81c2007cdd2deecb8293be8cb185e1a5b2 |
Pino Toscano <pino@debian.org> | not-needed | upstream | 2025-12-17 | |
| Change-XML-removeInvalidXml-to-check-chars-as-explic.diff | Change XML::removeInvalidXml() to check chars as explicitly signed The loading procedure of the Tellico own XML files has the following steps: (1) load the XML file as-is (2) in case (1) fails because of not well-formed XML: (2a) use XML::recoverFromBadXMLName() on the data (2b) in case (2a) did not change anything, use XML::removeInvalidXml() on the data (2c) in case the data changed in either (2a) or (2b), try to load the data . XML::recoverFromBadXMLName() (used in 2a) indirectly uses xmlValidateNCName() from libxml2, and it changed its behaviour: - before version 2.15: xmlValidateNCName() does not accept any non-ASCII name, so XML::recoverFromBadXMLName() results in only ASCII tags, and step (2b) is not done - starting from version 2.15: xmlValidateNCName() accepts Unicode names too, so the output of (2a) is unchanged, and thus (2b) is done; XML::removeInvalidXml() filters characters by checking them as 'char', and on architectures where char is unsigned (e.g. ARM) the existing check accepts also Unicode control characters, so the output is not changed either . The above is the flow when trying to load the bug418067.xml test case. . Since the step (2b) is a fallback for a fallback, and generally very much not likely to happen in the real world, then "freeze" the existing behaviour (i.e. the one happening on architectures with signed char, such as x86/x86_64). . Drive-by change: make 'c' as const. |
Pino Toscano <pino@kde.org> | yes | debian | 2025-12-18 |