Bug#1005201: boost1.74: outdated embedded data copy: unicode-data

[Paul Wise]

Source: boost1.74
Severity: normal
Usertags: embed

Unicode 14 was released and unicode-data 14.0.0-1.1 was uploaded.

The boost1.74 source package contains very very outdated embedded
data copies of several files from Unicode 5.2.0:

   $ head -n1 libs/spirit/workbench/unicode/*.txt
   ==> libs/spirit/workbench/unicode/DerivedCoreProperties.txt <==
   # DerivedCoreProperties-5.2.0.txt
  
   ==> libs/spirit/workbench/unicode/PropList.txt <==
   # PropList-5.2.0.txt
  
   ==> libs/spirit/workbench/unicode/Scripts.txt <==
   # Scripts-5.2.0.txt
  
   ==> libs/spirit/workbench/unicode/UnicodeData.txt <==
   0000;<control>;Cc;0;BN;;;;;N;NULL;;;;

Upstream appears to have converted those files into these files using
before releasing the tarball.

   libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/script_table.hpp
   libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/uppercase_table.hpp
   libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/category_table.hpp
   libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/lowercase_table.hpp
   libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/query.hpp

Most of these were generated using the create_tables code, which is in
the source package but is not built and run during the build process.
  
   $ grep -r create_tables
   debian/copyright: libs/spirit/workbench/unicode/create_tables.cpp
   $ grep -C5 AUTO libs/spirit/workbench/unicode/create_tables.cpp
           << "    Copyright (c) 2001-2011 Joel de Guzman\n"
           << "\n"
           << "    Distributed under the Boost Software License, Version 1.0. (See accompanying\n"
           << "    file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)\n"
           << "\n"
           << "    AUTOGENERATED. DO NOT EDIT!!!\n"
           << "==============================================================================*/\n"
           << "#include <boost/cstdint.hpp>\n"
           << "\n"
           << "namespace boost { namespace spirit { namespace ucd { namespace detail\n"
           << "{"
$ head -n8 libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/script_table.hpp
/*=============================================================================
Copyright (c) 2001-2011 Joel de Guzman

Distributed under the Boost Software License, Version 1.0. (See accompanying
file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)

AUTOGENERATED. DO NOT EDIT!!!
==============================================================================*/

One of those files (query.hpp) is generated by a python script instead
and that is not available in the source package, I think this is
therefore a severity serious bug but I'm not sure.

$ head -n9 libs/spirit/include/boost/spirit/home/support/char_encoding/unicode/query.hpp
/*=============================================================================
Copyright (c) 2001-2011 Joel de Guzman

Distributed under the Boost Software License, Version 1.0. (See accompanying
file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)

Autogenerated by MultiStageTable.py (Unicode multi-stage
table builder) (c) Peter Kankowski, 2008
==============================================================================*/

These files are then shipped in the libboost1.74-dev binary package.

Please ask upstream to remove the above files from the source package
and instead build-depend on the unicode-data binary package and convert
the relevant files to headers at build time.

--
bye,
pabs

https://wiki.debian.org/PaulWise

[Giovanni Mascellani]

Hi,

thanks for reopening the bug on the new Boost package. I agree that
these issues should be solved; unfortunately I am very low on time to do
that.

Thanks, Giovanni.


Il 09/02/22 00:25, Paul Wise ha scritto: