Debian Patches

Status for libur-perl/0.470+ds-4

Patch Description Author Forwarded Bugs Origin Last update
Honour-SOURCE_DATE_EPOCH-for-embedded-timestamp-in-genera.patch Honour SOURCE_DATE_EPOCH for embedded timestamp in generated documentation.

When the environment variable SOURCE_DATE_EPOCH is set, use it instead of the
current time for the embedded build timestamp.

The specification of SOURCE_DATE_EPOCH is available here:
https://reproducible-builds.org/specs/source-date-epoch/

In the case it is not defined, fall back to current time.
intrigeri <intrigeri@boum.org> no 2016-06-27
sqlite-case-insensitve.patch Fix test failure with newer (DBD::)SQLite ```
t/URT/t/04a_sqlite.t
…
not ok 80 - column details for table inline are correct
# Failed test 'column details for table inline are correct'
# at t/URT/t/04a_sqlite.t line 53.
# Structures begin differing at:
# $got->[0]{DATA_TYPE} = 'INTEGER'
# $expected->[0]{DATA_TYPE} = 'integer'
# Looks like you failed 1 test of 80.
Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/80 subtests
```
Use Test::Deep's cmp_deeply to be able to check against a case-insenstive regexp.
gregor herrmann <gregoa@debian.org> yes upstream vendor 2022-01-17
Perl-5.40-Missing-comma-after-first-argument-to-return.patch fix (test) failure with perl 5.40 # Failed test 'use UR::All;'
# at t/URT/t/00_load.t line 11.
# Tried to use 'UR::All'.
# Error: Missing comma after first argument to return at /build/libur-perl-0.470+ds/lib/UR/Namespace/Command/Define/Datasource/Sqlite.pm line 41, near ");"
.
# Failed test 'use UR::Namespace::Command::Define::Datasource::Sqlite;'
# at t/URT/t/80_command_define_datasource.t line 29.
# Tried to use 'UR::Namespace::Command::Define::Datasource::Sqlite'.
# Error: Missing comma after first argument to return at /build/libur-perl-0.470+ds/lib/UR/Namespace/Command/Define/Datasource/Sqlite.pm line 41, near ");"
gregor herrmann <gregoa@debian.org> yes upstream vendor 2024-06-16
fix-invalid-unicode-entities.patch Fix invalid unicode entities This patch fixes a test failure triggered by a conversion to XML that gives unexpected results:
.
-<object type="Acme::Cat::Set" id="Acme::Cat/And/owner_id/O:O:111">
+<object type="Acme::Cat::Set" id="Acme::Cat/And/owner_id/O:&#xFFFD;O:111&#xFFFD;">
.
An analysis done by IA indicates:
.
+++++++
.
The test failure in t/URT/t/63c_view_with_subviews.t:151-155 is caused by a behavioral change in libxml2 (>= 2.13.0) when serializing characters that are forbidden by the
XML 1.0 specification.
──────
### Root Cause Details
.
1. Control characters in Set IDs:
The failing subtest is rendering an XML view of $cat_set (Acme::Cat::Set). The id of a UR::Object::Set is the ID of its underlying UR::BoolExpr, which uses ASCII control
characters as delimiters:
• chr(29) (0x1D, Group Separator, defined in $UR::BoolExpr::Util::id_sep)
• chr(30) (0x1E, Record Separator, defined in $UR::BoolExpr::Util::record_sep)
Thus, $cat_set->id is:
"Acme::Cat/And/owner_id/O:\x1dO:111\x1e"
2. XML serialization in UR::Object::View::Default::Xml:
In Xml.pm:
$object->addChild( $xml_doc->createAttribute('id', $subject->id ) );
...
my $doc_string = $xml_doc->toString(1);
.
# remove invalid XML entities
$doc_string =~ s/[^\x09\x0A\x0D\x20-\x{D7FF}\x{E000}-\x{FFFD}\x{10000}-\x{10FFFF}]//go;
3. Behavior change in libxml2:
• In older libxml2 (< 2.13.0):
$xml_doc->toString(1) dumped the raw \x1d and \x1e bytes directly into $doc_string. The regex on line 105 then stripped those raw bytes out, leaving:
<object type="Acme::Cat::Set" id="Acme::Cat/And/owner_id/O:O:111">
This stripped string was saved as the reference fixture in t/URT/t/63c_view_with_subviews.t.expected.cat_set.xml:2.
• In newer libxml2 (>= 2.13.0, including 2.15.4):
During serialization, libxml2 now automatically substitutes illegal XML 1.0 characters with the Unicode replacement character entity (�).
As a result:
• $xml_doc->toString(1) returns id="Acme::Cat/And/owner_id/O:�O:111�".
• The subsequent regex does not touch � because it is composed entirely of valid ASCII characters (&, #, x, F, D, ;).
• Test 27 fails comparing got against expected:
- <object type="Acme::Cat::Set" id="Acme::Cat/And/owner_id/O:O:111">
+ <object type="Acme::Cat::Set" id="Acme::Cat/And/owner_id/O:&#xFFFD;O:111&#xFFFD;">
.
──────
### Possible Fix
.
Sanitize strings (removing invalid XML characters) before inserting them into DOM nodes/attributes in lib/UR/Object/View/Default/Xml.pm, rather than attempting to clean
them after toString():
.
my $id = $subject->id;
$id =~ s/[^\x09\x0A\x0D\x20-\x{D7FF}\x{E000}-\x{FFFD}\x{10000}-\x{10FFFF}]//go if defined $id;
$object->addChild( $xml_doc->createAttribute('id', $id ) );
.
++++
.
The analysis is good, but the solution is lackluster. I pushed the IA further by asking:
.
> are there other special characters beside char(29) and char(30) that are used by UR ?
.
++++
Yes. In lib/UR/BoolExpr/Util.pm:18-24, UR defines a full set of ASCII control characters used to serialize rule definitions and value IDs:
.
### Control characters in UR::BoolExpr::Util
.
Character │ Hex │ ASCII Name │ Variable Name │ Purpose in UR
───────────────────────────┼──────────────────────────┼────────────────────────────┼──────────────────────────┼─────────────────────────────────────────────────────────────
chr(20) │ 0x14 │ DC4 (Device Control 4) │ $empty_list │ Represents an empty arrayref []
chr(21) │ 0x15 │ NAK (Negative Acknowledge) │ $null_value │ Represents undef / SQL NULL
chr(28) │ 0x1C │ FS (File Separator) │ $empty_string │ Represents an empty string ""
chr(29) │ 0x1D │ GS (Group Separator) │ $id_sep │ Separates rule template ID from value ID
chr(30) │ 0x1E │ RS (Record Separator) │ $record_sep │ Delimits distinct property values within a value_id
chr(31) │ 0x1F │ US (Unit Separator) │ $unit_sep │ Delimits items inside a multi-value list (e.g. in => [...])
.
│ Important
│ All 6 of these characters fall into the ASCII C0 control range (0x00–0x1F) and are strictly forbidden in XML 1.0. If an object ID or property (such as a UR::BoolExpr or
│ UR::Object::Set) contains any of these (e.g., representing undef, empty strings, or lists), newer libxml2 will replace them with �.
──────
### Other Delimiters Used Across UR
.
1. "\t" (chr(9), Tab):
• Default composite_id_separator used to join multi-property primary keys in UR classes and UR::Object::Index. (Valid in XML 1.0).
2. "\0" (chr(0), NUL byte):
• Used in Command::Dispatch::Shell and Command::V1 to join command line arguments (join("\0", $0, @argv)). (Forbidden in XML 1.0).
3. "/" (Slash):
• Used as the composite_id_separator in UR::BoolExpr::Template (e.g., Acme::Cat/And/owner_id/O:).
4. "\n" (chr(10), Line Feed):
• Default record_separator for file-based datasources (UR::DataSource::File, UR::DataSource::Filesystem).
++++
.
All in all, this patch removes the control chars used by UR which are forbidden in XML from the `id` field before injecting data in XML::LibXML.
dod@debian no debian

All known versions for source package 'libur-perl'

Links