PR#6: Correctly set the soversion based on SemVer properties and split shared libraries into -libs - rpms/tesseract

rpms / tesseract

#6 Correctly set the soversion based on SemVer properties and split shared libraries into -libs

Merged 9 months ago by smani. Opened 9 months ago by salimma.

rpms/ salimma/tesseract rawhide-fix-soversion into rawhide

Split shared libraries into their own -libs subpackage

Michel Lind • 9 months ago

ffbb579

Correctly set the soversion based on SemVer properties

Michel Lind • 9 months ago

bc8aa11

tesseract.spec

file modified

+33 -8

		`@@ -8,18 +8,21 @@`

		`Name: tesseract`
		`Version: 5.4.1`
		`- Release: 2%{?dist}`
		`+ Release: 3%{?dist}`
		`Summary: Raw OCR Engine`

		`License: Apache-2.0`
		`URL: https://github.com/tesseract-ocr/%{name}`
		`Source0: https://github.com/tesseract-ocr/tesseract/archive/%{version}%{?pre:-%pre}/%{name}-%{version}%{?pre:-%pre}.tar.gz`

		`+ # Correctly set the soversion based on SemVer properties`
		`+ # backported from https://github.com/tesseract-ocr/tesseract/pull/4319.patch`
		`+ Patch0: tesseract_fix_soversion.patch`
		`# Fix library name case`
		`# Build training libs statically`
		`- Patch0: tesseract_cmake.patch`
		`+ Patch1: tesseract_cmake.patch`
		`# Don't assume neon available on arm64/aarch64`
		`- Patch1: tesseract_neon.patch`
		`+ Patch2: tesseract_neon.patch`


		`BuildRequires: cmake`
		`@@ -64,29 +67,43 @@`
		`BuildRequires: mingw64-pango`
		`%endif`

		`+ Requires: %{name}-libs%{?_isa} = %{version}-%{release}`
		`Requires: tesseract-langpack-eng`


		`- %description`
		`+ %global _description %{expand:`
		`A commercial quality OCR engine originally developed at HP between 1985 and`
		`1995. In 1995, this engine was among the top 3 evaluated by UNLV. It was`
		`- open-sourced by HP and UNLV in 2005.`
		`+ open-sourced by HP and UNLV in 2005.}`
		`+`
		`+ %description %_description`


		`%package devel`
		`Summary: Development files for %{name}`
		`Requires: %{name}%{?_isa} = %{version}-%{release}`

		`- %description devel`
		`+ %description devel %_description`
		`+`
		`The %{name}-devel package contains header file for`
		`developing applications that use %{name}.`


		`+ %package libs`
		`+ Summary: Shared libraries for %{name}`
		`+`
		`+ %description libs %_description`
		`+`
		`+ The %{name}-libs package contains shared libraries`
		`+ for %{name}.`
		`+`
		`+`
		`%package tools`
		`Summary: Training tools for %{name}`
		`Requires: %{name}%{?_isa} = %{version}-%{release}`

		`- %description tools`
		`+ %description tools %_description`
		`+`
		`The %{name}-tools package contains tools for training %{name}.`


		`@@ -167,7 +184,6 @@`
		`%license LICENSE`
		`%doc AUTHORS ChangeLog README.md`
		`%{_bindir}/%{name}`
		`- %{_libdir}/lib%{name}.so.5.4.1`
		`%{_datadir}/%{name}/`
		`%{_mandir}/man1/tesseract.1*`

		`@@ -179,6 +195,10 @@`
		`%{_libdir}/cmake/%{name}/`
		`%{_libdir}/pkgconfig/%{name}.pc`

		`+ %files libs`
		`+ %{_libdir}/lib%{name}.so.5.4`
		`+ %{_libdir}/lib%{name}.so.%{version}`
		`+`
		`%files tools`
		`%{_bindir}/ambiguous_words`
		`%{_bindir}/classifier_tester`
		`@@ -245,6 +265,11 @@`


		`%changelog`
		`+ * Mon Sep 23 2024 Michel Lind <salimma@fedoraproject.org> - 5.4.1-3`
		`+ - Correctly set the soversion based on SemVer properties`
		`+ Backport of upstream PR#4319 from Neal Gompa (ngompa)`
		`+ - Split shared libraries into their own -libs subpackage`
		`+`
		`* Sat Jul 20 2024 Fedora Release Engineering <releng@fedoraproject.org> - 5.4.1-2`
		`- Rebuilt for https://fedoraproject.org/wiki/Fedora_41_Mass_Rebuild`

tesseract_cmake.patch

file modified

+5 -7

		`@@ -1,6 +1,5 @@`
		`- diff -rupN --no-dereference tesseract-5.4.1/CMakeLists.txt tesseract-5.4.1-new/CMakeLists.txt`
		`- --- tesseract-5.4.1/CMakeLists.txt 2024-06-11 20:18:21.000000000 +0200`
		`- +++ tesseract-5.4.1-new/CMakeLists.txt 2024-06-13 18:47:53.277068523 +0200`
		`+ --- a/CMakeLists.txt`
		`+ +++ b/CMakeLists.txt`
		`@@ -364,7 +364,7 @@ elseif(UNIX)`
		`set(LIB_pthread pthread)`
		`endif()`
		`@@ -12,7 +11,7 @@`
		`add_definitions("-DCMAKE_BUILD")`
		`@@ -831,12 +831,17 @@ set_target_properties(`
		`libtesseract PROPERTIES SOVERSION`
		`- ${VERSION_MAJOR}.${VERSION_MINOR}.${VERSION_PATCH})`
		`+ ${VERSION_MAJOR}.${VERSION_MINOR})`

		`-set_target_properties(`
		`- libtesseract`
		`@@ -46,9 +45,8 @@`
		`endif()`

		`# ##############################################################################`
		`- diff -rupN --no-dereference tesseract-5.4.1/src/training/CMakeLists.txt tesseract-5.4.1-new/src/training/CMakeLists.txt`
		`- --- tesseract-5.4.1/src/training/CMakeLists.txt 2024-06-11 20:18:21.000000000 +0200`
		`- +++ tesseract-5.4.1-new/src/training/CMakeLists.txt 2024-06-13 18:47:53.277068523 +0200`
		`+ --- a/src/training/CMakeLists.txt`
		`+ +++ b/src/training/CMakeLists.txt`
		`@@ -115,7 +115,7 @@ if(NOT DISABLED_LEGACY_ENGINE)`
		`common/trainingsampleset.h)`
		`endif()`

tesseract_fix_soversion.patch

file added

+34

		`@@ -0,0 +1,34 @@`
		`+ From 280779c615402e335e40d96b9cf71ee8758fd63f Mon Sep 17 00:00:00 2001`
		`+ From: Neal Gompa <neal@gompa.dev>`
		`+ Date: Wed, 18 Sep 2024 07:44:29 -0400`
		`+ Subject: [PATCH] cmake: Correctly set the soversion based on SemVer properties`
		`+`
		`+ As this project follows Semantic Versioning, the shared object`
		`+ version should match these semantics.`
		`+`
		`+ The two options that make sense here are to have the soversion`
		`+ set to the version major (so only breaking changes are tracked)`
		`+ or to set to version major and minor (so breaking and API additions`
		`+ are tracked).`
		`+`
		`+ Since the Windows version of the library already uses version major`
		`+ and version minor, let's just do this universally.`
		`+`
		`+ Fixes: 832926f5afebd9a5a81343425e733708332f730f ("Update library version handling for cmake")`
		`+`
		`+ Signed-off-by: Neal Gompa <neal@gompa.dev>`
		`+ ---`
		`+ CMakeLists.txt \| 2 +-`
		`+ 1 file changed, 1 insertion(+), 1 deletion(-)`
		`+`
		`+ --- a/CMakeLists.txt`
		`+ +++ b/CMakeLists.txt`
		`+ @@ -829,7 +829,7 @@ set_target_properties(`
		`+ ${VERSION_MAJOR}.${VERSION_MINOR}.${VERSION_PATCH})`
		`+ set_target_properties(`
		`+ libtesseract PROPERTIES SOVERSION`
		`+ - ${VERSION_MAJOR}.${VERSION_MINOR}.${VERSION_PATCH})`
		`+ + ${VERSION_MAJOR}.${VERSION_MINOR})`
		`+`
		`+ set_target_properties(`
		`+ libtesseract`

salimma commented 9 months ago

I'm looking into making Tesseract packaging more modular so it's easier to backport newer Tesseracts in EPEL without affecting packages that depend on the version shipped by EL/CentOS Stream; in doing so @ngompa and I noticed that the soversion of Tesseract is overly strict (including the patch version, which it should not, and does not do on Windows)

This was fixed in https://github.com/tesseract-ocr/tesseract/pull/4319; this PR backports it and also splits off -libs.

Let me know if changes are needed, i.e. if tesseract dependents also need data files and therefore splitting libs no longer make sense (or they can be moved to a -data subpackage).

ngompa commented 9 months ago

LGTM.

smani commented 9 months ago

Thanks!

Pull-Request has been merged by smani

9 months ago

salimma commented 9 months ago

per https://bodhi.fedoraproject.org/updates/FEDORA-2024-41b92bf172#comment-3740912 - I forgot to mention in the PR - since now the soversion is just maj.min we do need to rebuild the dependents.

Rebuilding it in the ffmpeg sidetag

michel in tesseract on  rawhide [?] 
⬢ [f40-packaging] ❯ fedrq-whatrequires-verbose.sh 'libtesseract.so.5.4.1()(64bit)' -b f41
+ fedrq whatrequires 'libtesseract.so.5.4.1()(64bit)' -F multiline:source,requires -b f41
+ grep 'libtesseract.so.5.4.1()(64bit)'
R-tesseract : libtesseract.so.5.4.1()(64bit)
crow-translate : libtesseract.so.5.4.1()(64bit)
gimagereader : libtesseract.so.5.4.1()(64bit)
gimagereader : libtesseract.so.5.4.1()(64bit)
ffmpeg : libtesseract.so.5.4.1()(64bit)
mupdf : libtesseract.so.5.4.1()(64bit)
opencv : libtesseract.so.5.4.1()(64bit)
skanpage : libtesseract.so.5.4.1()(64bit)
tesseract : libtesseract.so.5.4.1()(64bit)
tesseract : libtesseract.so.5.4.1()(64bit)

michel in tesseract on  rawhide [?] 
⬢ [f40-packaging] ❯ fedrq-whatrequires-verbose.sh 'libtesseract.so.5.4.1()(64bit)' -b f42
+ fedrq whatrequires 'libtesseract.so.5.4.1()(64bit)' -F multiline:source,requires -b f42
+ grep 'libtesseract.so.5.4.1()(64bit)'
R-tesseract : libtesseract.so.5.4.1()(64bit)
R-tesseract : libtesseract.so.5.4.1()(64bit)
crow-translate : libtesseract.so.5.4.1()(64bit)
crow-translate : libtesseract.so.5.4.1()(64bit)
gimagereader : libtesseract.so.5.4.1()(64bit)
gimagereader : libtesseract.so.5.4.1()(64bit)
gimagereader : libtesseract.so.5.4.1()(64bit)
gimagereader : libtesseract.so.5.4.1()(64bit)
ffmpeg : libtesseract.so.5.4.1()(64bit)
ffmpeg : libtesseract.so.5.4.1()(64bit)
maui-mauikit-imagetools : libtesseract.so.5.4.1()(64bit)
maui-mauikit-imagetools : libtesseract.so.5.4.1()(64bit)
mupdf : libtesseract.so.5.4.1()(64bit)
mupdf : libtesseract.so.5.4.1()(64bit)
opencv : libtesseract.so.5.4.1()(64bit)
opencv : libtesseract.so.5.4.1()(64bit)
skanpage : libtesseract.so.5.4.1()(64bit)
skanpage : libtesseract.so.5.4.1()(64bit)
tesseract : libtesseract.so.5.4.1()(64bit)
tesseract : libtesseract.so.5.4.1()(64bit)
tesseract : libtesseract.so.5.4.1()(64bit)
tesseract : libtesseract.so.5.4.1()(64bit)