SBo scripts not building on current (read 1st post, pls)

ArTourter · 11-09-2020, 04:29 AM

python3-PyYAML doesn't build for me on -current with python3.9. Simply bumping the version to 5.3.1 fixes the problem here.

Hope this helps

ponce · 11-09-2020, 04:59 AM

Quote:

Originally Posted by ArTourter

python3-PyYAML doesn't build for me on -current with python3.9. Simply bumping the version to 5.3.1 fixes the problem here.

Hope this helps

thanks, Greg!

USUARIONUEVO · 11-12-2020, 04:20 PM

Attention in todays perl update , including new modules ...all present in SBo and now need to remove it.

Quote:

Added: Module-Runtime-0.016, Moo-2.004000, Path-Tiny-0.114,
Sub-Quote-2.006006, Template-Toolkit-3.009.

Thanks!

ponce · 11-13-2020, 12:33 AM

Quote:

Originally Posted by USUARIONUEVO

Attention in todays perl update , including new modules ...all present in SBo and now need to remove it.

Thanks!

thanks for the reminder but I had already removed them

M0M0 · 11-13-2020, 03:14 AM

Quote:

Originally Posted by montagdude

webkit2gtk needs this patch to the SlackBuild:

Code:

diff --git a/libraries/webkit2gtk/webkit2gtk.SlackBuild b/libraries/webkit2gtk/webkit2gtk.SlackBuild
old mode 100644
new mode 100755
index dac584b438..777c40c9a4
--- a/libraries/webkit2gtk/webkit2gtk.SlackBuild
+++ b/libraries/webkit2gtk/webkit2gtk.SlackBuild
@@ -55,6 +55,9 @@ else
   LIBDIRSUFFIX=""
 fi
 
+# Fix for icu4c change
+SLKCFLAGS="$SLKCFLAGS -DU_DEFINE_FALSE_AND_TRUE=1"
+
 # This is needed to fix build issue on single make job
 # Thanks to ponce for investigating this issue.
 export MAKEFLAGS=${MAKEFLAGS:--j2}
@@ -77,6 +80,8 @@ find -L . \
 mkdir -p build
 cd build
   cmake -DPORT=GTK \
+	-DCMAKE_C_FLAGS:STRING="$SLKCFLAGS" \
+	-DCMAKE_CXX_FLAGS:STRING="$SLKCFLAGS" \
 	-DCMAKE_BUILD_TYPE=Release \
         -DCMAKE_SKIP_RPATH=ON \
 	-DCMAKE_INSTALL_PREFIX=/usr \

See this thread:
https://www.linuxquestions.org/quest...nt-4175684551/

That's probably a stupid question, but how do you apply this patch? When I run this code while in the directory that contains Slackbuild, I see differences in the .Slackbuild file, but when I try to compile the program I still get errors. Should I do something else with this patch, like inserting it into a file, or is my problem a different one?

ponce · 11-13-2020, 04:00 AM

Quote:

Originally Posted by M0M0

That's probably a stupid question, but how do you apply this patch? When I run this code while in the directory that contains Slackbuild, I see differences in the .Slackbuild file, but when I try to compile the program I still get errors. Should I do something else with this patch, like inserting it into a file, or is my problem a different one?

here you can find the latest version of the script tested on the latest current that will go in the repository linked in the first post at the next merge

http://ponce.cc/slackware/testing/webkit2gtk/

avian · 11-14-2020, 02:30 AM

When building package x265 (v3.4) on a fully upgraded -current system, it seems to report the version as 0.0 -

Code:

# x265 -V
x265 [info]: HEVC encoder version 0.0
x265 [info]: build info [Linux][GCC 9.3.0][64 bit] 8bit+10bit+12bit
x265 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2

and also in /usr/lib64/pkgconfig/x265.pc

Code:

# cat /usr/lib64/pkgconfig/x265.pc 
prefix=/usr
exec_prefix=${prefix}
libdir=${exec_prefix}/lib64
includedir=${prefix}/include

Name: x265
Description: H.265/HEVC video encoder
Version: 0.0
Libs: -L${libdir} -lx265
Libs.private: -lstdc++ -lm -lgcc_s -lgcc -lgcc_s -lgcc -lrt -ldl
Cflags: -I${includedir}

Didnt have this problem using the 3.2 build script with 3.2 source on -current. It seems version.cmake defaults to 0.0 if the script cant find anything better, but its way above my patch-fu level to create a patch that isnt remotely hacky and ugly.

ponce · 11-14-2020, 03:12 AM

Quote:

Originally Posted by avian

When building package x265 (v3.4) on a fully upgraded -current system, it seems to report the version as 0.0 -

Code:

# x265 -V
x265 [info]: HEVC encoder version 0.0
x265 [info]: build info [Linux][GCC 9.3.0][64 bit] 8bit+10bit+12bit
x265 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2

and also in /usr/lib64/pkgconfig/x265.pc

Code:

# cat /usr/lib64/pkgconfig/x265.pc 
prefix=/usr
exec_prefix=${prefix}
libdir=${exec_prefix}/lib64
includedir=${prefix}/include

Name: x265
Description: H.265/HEVC video encoder
Version: 0.0
Libs: -L${libdir} -lx265
Libs.private: -lstdc++ -lm -lgcc_s -lgcc -lgcc_s -lgcc -lrt -ldl
Cflags: -I${includedir}

Didnt have this problem using the 3.2 build script with 3.2 source on -current. It seems version.cmake defaults to 0.0 if the script cant find anything better, but its way above my patch-fu level to create a patch that isnt remotely hacky and ugly.

hi avian, the one you describe is not a problem related to current as the build system behaves exactly the same on 14.2 and the version reported by "x265 -V" is 0.0 also there: I suggest you to report the issue upstream to the x265 developers.
if you want to fix it in the meantime you can add this line just after the find block in the SlackBuild

Code:

 sed -i "s|set(X265_LATEST_TAG\ \"0.0\")|set(X265_LATEST_TAG \"$VERSION\")|" source/cmake/version.cmake

avian · 11-14-2020, 07:13 AM

Quote:

Originally Posted by ponce

hi avian, the one you describe is not a problem related to current as the build system behaves exactly the same on 14.2 and the version reported by "x265 -V" is 0.0 also there: I suggest you to report the issue upstream to the x265 developers.
if you want to fix it in the meantime you can add this line just after the find block in the SlackBuild

Code:

 sed -i "s|set(X265_LATEST_TAG\ \"0.0\")|set(X265_LATEST_TAG \"$VERSION\")|" source/cmake/version.cmake

Thanks Ponce. Your temporary fix seems perfect. I'll use that until the issue is fixed upstream.

sombragris · 11-14-2020, 04:26 PM

EDIT: Removed. The issue seems to be fixed by now. Nevermind...

montagdude · 11-20-2020, 08:59 AM

Quote:

Originally Posted by ponce

thanks, Greg!

Similarly with scipy3:

Code:

scipy/cluster/_hierarchy.c: In function ‘__Pyx_modinit_type_init_code’:
scipy/cluster/_hierarchy.c:34720:47: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34720 |   __pyx_type_5scipy_7cluster_10_hierarchy_Heap.tp_print = 0;
      |                                               ^
scipy/cluster/_hierarchy.c:34732:59: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34732 |   __pyx_type_5scipy_7cluster_10_hierarchy_LinkageUnionFind.tp_print = 0;
      |                                                           ^
scipy/cluster/_hierarchy.c:34741:100: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34741 | double____double____double____double____int____int____int___to_py.tp_print = 0;
      |                                                                  ^

scipy/cluster/_hierarchy.c:34749:25: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34749 |   __pyx_type___pyx_array.tp_print = 0;
      |                         ^
scipy/cluster/_hierarchy.c:34754:31: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34754 |   __pyx_type___pyx_MemviewEnum.tp_print = 0;
      |                               ^
scipy/cluster/_hierarchy.c:34769:30: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34769 |   __pyx_type___pyx_memoryview.tp_print = 0;
      |                              ^
scipy/cluster/_hierarchy.c:34782:35: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34782 |   __pyx_type___pyx_memoryviewslice.tp_print = 0;
      |                                   ^

Upgrading to scipy-1.5.4 fixes it.

ponce · 11-20-2020, 10:49 AM

Quote:

Originally Posted by montagdude

Similarly with scipy3:

Code:

scipy/cluster/_hierarchy.c: In function ‘__Pyx_modinit_type_init_code’:
scipy/cluster/_hierarchy.c:34720:47: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34720 |   __pyx_type_5scipy_7cluster_10_hierarchy_Heap.tp_print = 0;
      |                                               ^
scipy/cluster/_hierarchy.c:34732:59: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34732 |   __pyx_type_5scipy_7cluster_10_hierarchy_LinkageUnionFind.tp_print = 0;
      |                                                           ^
scipy/cluster/_hierarchy.c:34741:100: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34741 | double____double____double____double____int____int____int___to_py.tp_print = 0;
      |                                                                  ^

scipy/cluster/_hierarchy.c:34749:25: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34749 |   __pyx_type___pyx_array.tp_print = 0;
      |                         ^
scipy/cluster/_hierarchy.c:34754:31: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34754 |   __pyx_type___pyx_MemviewEnum.tp_print = 0;
      |                               ^
scipy/cluster/_hierarchy.c:34769:30: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34769 |   __pyx_type___pyx_memoryview.tp_print = 0;
      |                              ^
scipy/cluster/_hierarchy.c:34782:35: error: ‘PyTypeObject’ {aka ‘struct _typeobject’} has no member named ‘tp_print’
34782 |   __pyx_type___pyx_memoryviewslice.tp_print = 0;
      |                                   ^

Upgrading to scipy-1.5.4 fixes it.

thanks, pushed!

USUARIONUEVO · 11-20-2020, 06:13 PM

Hi i found 2 packages break

snort
dsniff

the two with same problem , cause glibc removes /usr/include/rpc/rpc.h

I think more network packages are involved, i say if sound some more.

I try upgrade snort , but no lucky...dsniff seems to have a patch in archlinux , but fails in our sources , im not see why.

USUARIONUEVO · 11-20-2020, 06:13 PM

repeated sorry.

ponce · 11-21-2020, 02:35 AM

Quote:

Originally Posted by USUARIONUEVO

Hi i found 2 packages break

snort
dsniff

the two with same problem , cause glibc removes /usr/include/rpc/rpc.h

I think more network packages are involved, i say if sound some more.

I try upgrade snort , but no lucky...dsniff seems to have a patch in archlinux , but fails in our sources , im not see why.

thanks for the report: they are both fixed in their own branches and they will go in at the next merge.