From: Brad Smith Subject: Re: UPDATE: libass 0.17.3 To: ports@openbsd.org Date: Sun, 27 Oct 2024 18:38:44 -0400 ping. On 2024-10-03 12:29 a.m., Brad Smith wrote: > ping. > > On 2024-08-27 11:23 p.m., Brad Smith wrote: >> ping. >> >> On 2024-08-03 4:31 p.m., Brad Smith wrote: >>> Here is an update to libass 0.17.3. >>> >>> >>> libass (0.17.3) >>>   * Fix 0.17.2 regression in the fontconfig fontprovider >>>     leading to undesirable widths being chosen from large >>> typographic families >>>   * Fix configure generated with slibtool-provided autoconf macros >>>   * Fix "make check" for shared-only builds >>>   * Constify some API parameters in a backwards-compatible manner >>>   * Add new ass_malloc and ass_free API functions >>>   * Tweak default optimization flags >>>   * Speed up parsing of events with very long override blocks >>>   * Improve handling of HarfBuzz-related failures >>> >>> libass (0.17.2) >>>   * Fix rendering of \h in certain cases >>>   * Fix a minor memory leak in the CoreText and DirectWrite font >>> provider >>>   * Fix wrong ASS_Image dimensions for huge BorderStyle=4 backgrounds >>>     potentially leading to out of bound reads by API users >>> >>>   * Improve quality of animated rectangular clips >>>   * Improve accuracy of cache limits >>>   * Full-Unicode cmaps are now always preferred >>>   * Improve font selection compatibility in the DirectWrite font >>> provider >>>   * Improved documentation >>>     * Color mangling notes in ass_types.h were updated for tone >>> mapping and HDR video >>>     * Guidelines for producing portable ASS files can now be found >>> in our wiki >>> >>>   * Updating selective overrides now forces a cache clear to avoid >>> issues with outdated caches >>>   * Frame and storage resolutions are now limited to what a single >>> ASS_Image can represent >>>   * make check now runs checkasm if assembly is enabled >>>   * CoreText can now be used on Mac OS X 10.5 >>>   * Meson/muon is now offered as a secondary build system bringing >>> back first-party MSVC support >>>     Note however it is not at feature parity with autotools in all >>> cases >>> >>>   * aarch64: add optimized assembly routines covering the same set >>> as on x86; >>>     they work on both little and big endian systems >>>   * x86: add SSSE3 versions of some assembly routines to help CPUs >>> without AVX2 >>>   * x86: it is now possible to build binaries with optimized >>> assembly and SHSTK support >>> >>>   * Improve VSFilter compatibility: >>>     * Trailing spaces are no longer trimmed from style and >>>       event string fields (this does not affect Text) >>>     * All BiDi paragraph separators now delimit BiDi runs >>>     * Fix treatment of font weights < 10 >>>     * Match GDI's faux italic slant >>>     * Employ faux italics and faux bolding in the same cases as GDI >>>     * Better match font selection with regard to italic and weight >>>     * Treat empty Style name as "Default" >>>     * Various drawing fixes >>>       * Reject drawings not starting with a valid m command or >>> invalid m followed by valid n >>>       * Drop coordinates following invalid commands rather than >>> appending them to the prior command >>>       * Ignore p commands without enough preceding nodes >>>       * Partially fix s and p commands used outside of a properly >>> closed b-spline sequence >>> >>> >>> Index: multimedia/libass/Makefile >>> =================================================================== >>> RCS file: /cvs/ports/multimedia/libass/Makefile,v >>> retrieving revision 1.33 >>> diff -u -p -u -p -r1.33 Makefile >>> --- multimedia/libass/Makefile    27 Feb 2024 08:59:42 -0000 1.33 >>> +++ multimedia/libass/Makefile    3 Aug 2024 20:28:43 -0000 >>> @@ -1,13 +1,12 @@ >>>   COMMENT=    portable ASS/SSA subtitle renderer >>>   -VER=        0.17.1 >>> +VER=        0.17.3 >>>   DISTNAME=    libass-${VER} >>>   CATEGORIES=    multimedia devel >>>   SITES= https://github.com/libass/libass/releases/download/${VER}/ >>>   EXTRACT_SUFX=    .tar.xz >>> -REVISION=    0 >>>   -SHARED_LIBS=    ass    3.1 >>> +SHARED_LIBS=    ass    3.2 >>>     HOMEPAGE=    https://github.com/libass/libass >>>   Index: multimedia/libass/distinfo >>> =================================================================== >>> RCS file: /cvs/ports/multimedia/libass/distinfo,v >>> retrieving revision 1.21 >>> diff -u -p -u -p -r1.21 distinfo >>> --- multimedia/libass/distinfo    9 Apr 2023 06:14:31 -0000 1.21 >>> +++ multimedia/libass/distinfo    3 Aug 2024 20:28:43 -0000 >>> @@ -1,2 +1,2 @@ >>> -SHA256 (libass-0.17.1.tar.xz) = >>> 8NoLv7pHbBauPhz9hiJW0wkVkR96uqGxbOYu5lMZJ4Q= >>> -SIZE (libass-0.17.1.tar.xz) = 403160 >>> +SHA256 (libass-0.17.3.tar.xz) = >>> 6uQl2lDwAVwh97OpxyYqkQ8CGK9GniLikxRi/tPFCVk= >>> +SIZE (libass-0.17.3.tar.xz) = 442744 >>> Index: multimedia/libass/patches/patch-libass_x86_x86inc_asm >>> =================================================================== >>> RCS file: >>> /cvs/ports/multimedia/libass/patches/patch-libass_x86_x86inc_asm,v >>> retrieving revision 1.1 >>> diff -u -p -u -p -r1.1 patch-libass_x86_x86inc_asm >>> --- multimedia/libass/patches/patch-libass_x86_x86inc_asm 27 Feb >>> 2024 08:59:42 -0000    1.1 >>> +++ multimedia/libass/patches/patch-libass_x86_x86inc_asm    3 Aug >>> 2024 20:28:43 -0000 >>> @@ -3,7 +3,7 @@ add ENDBR64 instructions for BTI >>>   Index: libass/x86/x86inc.asm >>>   --- libass/x86/x86inc.asm.orig >>>   +++ libass/x86/x86inc.asm >>> -@@ -756,6 +756,9 @@ BRANCH_INSTR jz, je, jnz, jne, jl, jle, jnl, >>> jnle, jg, >>> +@@ -839,6 +839,9 @@ BRANCH_INSTR jz, je, jnz, jne, jl, jle, jnl, >>> jnle, jg, >>>        %endif >>>        align function_align >>>        %2: >>> @@ -13,7 +13,7 @@ Index: libass/x86/x86inc.asm >>>        RESET_MM_PERMUTATION        ; needed for x86-64, also makes >>> disassembly somewhat nicer >>>        %xdefine rstk rsp           ; copy of the original stack >>> pointer, used when greater alignment than the known stack alignment >>> is required >>>        %assign stack_offset 0      ; stack pointer offset relative >>> to the return address >>> -@@ -777,6 +780,9 @@ BRANCH_INSTR jz, je, jnz, jne, jl, jle, jnl, >>> jnle, jg, >>> +@@ -860,6 +863,9 @@ BRANCH_INSTR jz, je, jnz, jne, jl, jle, jnl, >>> jnle, jg, >>>            global current_function %+ %1 >>>        %endif >>>        %1: