harfbuzz/pango: update to latest releases

[mascguy]

Description

• harfbuzz: update to 2.9.0
• pango: update to 1.48.9
• py-pygtk: patch for latest pango

Closes:

• 61814 - pango 1.42.4: current stable release is 1.48.9

Credit goes to @i0ntempest, who did most of the work as part of PR 11772 - gimp2: update to 2.10.24

Type(s)

• bugfix
• enhancement
• security fix

Tested on

macOS 10.12.6 16G2136
Xcode 9.2 9C40b

macOS 10.13.6 17G14019
Xcode 10.1 10B61

macOS 10.14.6 18G103
Xcode 11.3.1 11C505

macOS 10.15.6 19G2021
Xcode 12.0 12A7209

macOS 10.15.7 19H1217
Xcode 12.0 12A7209

Verification

Have you

• followed our Commit Message Guidelines?
• squashed and minimized your commits?
• checked that there aren't other open pull requests for the same change?
• referenced existing tickets on Trac with full URL?
• checked your Portfile with port lint?
• tried existing tests with sudo port test?
• tried a full install with sudo port -vst install?
• tested basic functionality of all binary files?

[macportsbot]

Notifying maintainers:
@ryandesign for port harfbuzz, pango.

[evanmiller]

Attempting to build pango with your Portfile on a Tiger box I get a configure error:

Run-time dependency xrender found: YES 0.9.10
Run-time dependency appleframeworks found: NO (tried framework)

meson.build:373:2: ERROR: Dependency "appleframeworks" not found, tried framework

It's using macports-gcc-7. Any ideas?

[Gcenx]

Maybe add a note about why this is required, since this isn’t needed for newer versions of glib

[Gcenx]

Maybe add a note about why this is required, since this isn’t needed for newer versions of glib

[mascguy]

@i0ntempest Zhenfu, any thoughts?

[i0ntempest]

This might be a mistake that I forgot to remove, it was used to force meson to not build glib as a subproject. Since we now have an updated glib this should be able to be removed without problems.

[Gcenx]

@i0ntempest Isn't that also the case for -Wno-error,-Wimplicit-fallthrough as I’ve not needed that for my updated version of pango.

[i0ntempest]

That cflag was indeed needed to build on my system when I tried, did another version came out? If yes it might not be needed anymore.

[Gcenx]

I have glib-2.68.4 installed on my end and haven't needed this cflag on Mojave

[mascguy]

Attempting to build pango with your Portfile on a Tiger box I get a configure error:

Run-time dependency xrender found: YES 0.9.10
Run-time dependency appleframeworks found: NO (tried framework)

meson.build:373:2: ERROR: Dependency "appleframeworks" not found, tried framework

It's using macports-gcc-7. Any ideas?

Hey Evan, I'll happily try to help, once the overall changes are complete. (Still looking at a few things, based on the outstanding comments.)

In the interim, is there a lower-level Meson log with detail?

[evanmiller]

@mascguy There is a lower level Meson log, but it contains no other information about “appleframeworks” besides what I posted. I can try a verbose mode if you have any pointers.

[mascguy]

@mascguy There is a lower level Meson log, but it contains no other information about “appleframeworks” besides what I posted. I can try a verbose mode if you have any pointers.

This is the first time I've dealt with Meson before. So until I'm a bit more familiar with it, I won't be much help.

[Gcenx]

if host_system == 'darwin'
  has_core_text = cc.links('''#include <Carbon/Carbon.h>
                              int main (void) {
                                CTGetCoreTextVersion ();
                                return 0;
                              }''',
                           name: 'CoreText availability',
                           dependencies: dependency('appleframeworks', modules: 'ApplicationServices'))
  if has_core_text
    pango_conf.set('HAVE_CORE_TEXT', 1)
  endif

  pango_deps += dependency('appleframeworks', modules: [ 'CoreFoundation', 'ApplicationServices' ])
endif

So I’d guess it’s not finding ApplicationServices

[mascguy]

Update: We finally have a working pango-devel - based on this PR - after creating a corresponding Devel version of harfbuzz. (It seems that these two are more intertwined than first expected.)

So that's a sigh of relief.

However, we still need to test and validate components dependent on these two. And that will require updating some ports to use a path-style dependency on harfbuzz, to support using the Devel version of the latter. (Whether those changes are committed to master, vs. simply testing locally with them, is still TBD. And if anyone has any thoughts/opinions on that, please don't hesitate to chime in!)

[evanmiller]

One more issue with pango-devel and meson: the GObject Introspection dylib location is incorrect:

$ g-ir-inspect --print-shlibs Pango
shlib: ./pango/libpango-1.0.0.dylib

It should look like GLib's, with the full installed path instead of the relative build path:

$ g-ir-inspect --print-shlibs GLib
shlib: /opt/local/lib/libglib-2.0.0.dylib
shlib: /opt/local/lib/libgobject-2.0.0.dylib

I am working on a patch for Meson to address this.

[evanmiller]

The weeds get pretty thick with the GObject Introspection bug – which affects many, many ports – but #12209 should work around it for pango-devel and other Meson-based projects.

[evanmiller]

All of my PPC-related concerns have now been addressed 👍

[mascguy]

Observers: Since I'm now taking on co-maintainership of these ports - and in the interest of moving forward - I'm merging this. And I'll certainly take ownership of any fallout.

[evanmiller]

Maybe close https://trac.macports.org/ticket/63496 while you're at it

[i0ntempest]

Found something weird: the pygtk-2.0.pc file in revision 4 of py27-pygtk is broken, line 6 should have ${exec_prefix} but instead it has ${PYTHON_EXEC_PREFIX} which is undefined and causes GIMP build to fail.

[jmroot]

The variables set by automake aren't really designed to be used outside of the makefiles it generates, and this is the sort of thing that can happen if you do that. The new variable needs to be substituted in to the .pc file with a line like this in the input file:

PYTHON_EXEC_PREFIX=@PYTHON_EXEC_PREFIX@

[mascguy]

Thanks so much guys, appreciate the help!

My initial pass at fixing this, involves patching configure.ac. Partly because that's how a number of other Python ports do it, and using the 'PYTHON_EXEC_PREFIX' approach resulted in issues. (And simply need to fix this ASAP, to stop the bleeding.) But certainly open to a simpler/better approach, once the dust settles.

The commit is: 1a0234c

[mascguy]

And my apologies to anyone bitten by the issue with py-pygtk: Despite our best efforts, we somehow missed this. But this is all on me, as it wasn't included in Zhenfu's changeset for the GIMP update PR; it was a late addition by Yours Truly.