On Sun, 13 Sep 2026 16:39:10 +0200 Kirill A. Korinsky <kirill@korins.ky> wrote: > On Sun, 13 Sep 2026 13:09:36 +0200, > Nazarenko Mykyta <mn@superpc.suicide.solutions> wrote: > > > > On Sat, 12 Sep 2026 23:34:30 +0200 > > Kirill A. Korinsky <kirill@korins.ky> wrote: > > > > > On Sat, 12 Sep 2026 22:08:59 +0200, > > > Nazarenko Mykyta <mn@superpc.suicide.solutions> wrote: > > > > > > > > [1 <text/plain; UTF-8 (quoted-printable)>] > > > > On Sat, 12 Sep 2026 12:15:11 +0000 > > > > Klemens Nanni <kn@openbsd.org> wrote: > > > > > > > > > 12.09.2026 01:42, Nazarenko Mykyta пишет: > > > > > > Hola ports@, kirill@, kn@. > > > > > > > > > > > > 64Gram has been updated to version 1.2.8, synchronized with tdesktop 7.1.3. > > > > > > > > > > > > Here is the updated port that I suggest importing. > > > > > > > > > > > > Compared to the original tdesktop, 64Gram provides additional > > > > > > security and privacy controls, including screenshot mode and > > > > > > the ability to disable cloud synchronization of message drafts. > > > > > > > > > > > > It also exposes user IDs for consistent account identification across > > > > > > username or avatar changes, and supports exporting messages in > > > > > > JSON format. > > > > > > > > > > > > The full list of additional features is documented here : > > > > > > https://github.com/TDesktop-x64/tdesktop/blob/dev/features.md > > > > > > > > > > I'm not very keen another heavy telegram desktop port that will want > > > > > frequent commits, syncing with net/tdesktop and testing... > > > > > > > > > > But that'd be the maintainer's job not mine and we also have a bunch > > > > > of similar chrom* flavoured ports, so If others deem 64gram useful, > > > > > I don't object. > > > > > > > > > > > > > > > Port-wise this looks fine with the following diff to > > > > > a) drop unused gsl dep > > > > > b) reduce diff with tdesktop > > > > > c) usual style (empty lines between targets and most variables) > > > > > > > > > > > > Thanks :) > > > > > > > > > > > > With best regards, > > > > > > Nazarenko Mykyta > > > > > > > > > > diff --git a/net/64gram/Makefile b/net/64gram/Makefile > > > > > index 4de2eeb8d..e443c7d78 100644 > > > > > --- a/net/64gram/Makefile > > > > > +++ b/net/64gram/Makefile > > > > > @@ -52,7 +52,6 @@ BUILD_DEPENDS = devel/boost,-main \ > > > > > devel/crc32c \ > > > > > devel/fmt \ > > > > > devel/gobject-introspection \ > > > > > - devel/microsoft-gsl \ > > > > > devel/range-v3>=0.11.0p0 \ > > > > > devel/tl-expected \ > > > > > net/tdlib/tde2e \ > > > > > @@ -86,11 +85,13 @@ RUN_DEPENDS = devel/desktop-file-utils \ > > > > > > > > > > # opt-out since 5.15, requires more bundled stuff like fcitx5-qt, hime, nimf > > > > > CONFIGURE_ARGS += -DDESKTOP_APP_DISABLE_QT_PLUGINS=ON > > > > > + > > > > > # Follow suit with most other distribution packages, i.e. use the token from > > > > > # Telegram's official Snap package > > > > > # https://github.com/telegramdesktop/tdesktop/issues/17435 > > > > > CONFIGURE_ARGS += -DTDESKTOP_API_ID=611335 \ > > > > > -DTDESKTOP_API_HASH=d524b414d21f4d37f08684c1df41ac9c > > > > > + > > > > > # Since cmake 3.25.1, BSDs are no longer recognised as Linux (in this port). > > > > > # Upstream should probably become BSD aware, but until that happens, > > > > > # pretend we're Linux to keep building... > > > > > @@ -100,6 +101,7 @@ NO_TEST = Yes > > > > > > > > > > # unported and required > > > > > NEEDED_BUNDLES = cld3 libprisma rlottie tgcalls MicroTeX cmark-gfm > > > > > + > > > > > # ported but CMake code hardcodes bundled file paths > > > > > NEEDED_BUNDLES += xdg-desktop-portal > > > > > > > > > > @@ -111,9 +113,10 @@ post-extract: > > > > > # All bundles: > > > > > @cd ${WRKSRC}/Telegram/ThirdParty/ && ls -x > > > > > # Needed bundles (others are removed to prevent accidential use): > > > > > - @cd ${WRKSRC}/Telegram/ThirdParty/ && mkdir -p .need/ && \ > > > > > + @cd ${WRKSRC}/Telegram/ThirdParty/ && mkdir .need/ && \ > > > > > mv ${NEEDED_BUNDLES} .need/ && rm -fr -- ./* && \ > > > > > mv .need/* . && rmdir .need/ && ls -x > > > > > + > > > > > post-patch: > > > > > # XXX fix the following on every source file > > > > > # warning: unknown warning option '-Wno-maybe-uninitialized' > > > > > @@ -126,6 +129,7 @@ post-patch: > > > > > # placed in cppgir.ignore > > > > > pre-configure: > > > > > ${INSTALL_SCRIPT} ${FILESDIR}/cppgir_clean.sh ${WRKSRC} > > > > > + > > > > > # reduce binary size fromm +660M to ~250M > > > > > do-install: > > > > > ${MODCMAKE_INSTALL_TARGET} --strip > > > > > > > > > > > > > Thanks kn@. > > > > > > > > New tarball attached with applied diff. > > > > > > > > > > I potentially interested in this port. > > > > > > So far I see ony one remaning question. > > > > > > Let assume that tomorrow new release tdesktop and 64gram is here, and it > > > needs new tdlib and tde2e. > > > > > > Am I right that it somehow compatible and it can be updated? > > > > > > Or one of ports ends with broken state on update and it should be updated at > > > the same time? > > > > > > If yes, that happens with time which 64gram team need to sync with their upstream? > > > > > > > I don't think anything will break, since in the ports-tree, tdlib isn't > > updated often enough for the ABI/API to break, 64gram doesn't catch up > > with tdesktop quite as quickly, but compatibility with the versions of > > the libraries used by tdesktop is ensured. > > > > I had tried a build in clean chroot, it fails as: > > In file included from /usr/ports/pobj/64gram-1.2.8/64Gram-1.2.8-full/Telegram/SourceFiles/webauthn/cable_scanner_linux.cpp:17: > In file included from /usr/ports/pobj/64gram-1.2.8/build-amd64/gen/bluez/bluez.hpp:9: > In file included from /usr/ports/pobj/64gram-1.2.8/build-amd64/gen/gio/gio.hpp:9: > In file included from /usr/ports/pobj/64gram-1.2.8/build-amd64/gen/glib/glib.hpp:34: > /usr/ports/pobj/64gram-1.2.8/build-amd64/gen/glib/_constants.hpp:26:46: error: use of undeclared identifier 'G_C_STD_VERSION'; did you mean 'C_STD_VERSION_'? > 26 | GI_MODULE_INLINE const gint C_STD_VERSION_ = G_C_STD_VERSION; > | ^~~~~~~~~~~~~~~ > /usr/ports/pobj/64gram-1.2.8/build-amd64/gen/glib/_constants.hpp:26:29: note: 'C_STD_VERSION_' declared here > 26 | GI_MODULE_INLINE const gint C_STD_VERSION_ = G_C_STD_VERSION; > | ^ > /usr/ports/pobj/64gram-1.2.8/build-amd64/gen/glib/_constants.hpp:148:30: error: cannot initialize a variable of type 'const gchar' (aka 'const char') with an > rvalue of type 'gchar *' (aka 'char *') > 148 | GI_MODULE_INLINE const gchar LOG_DOMAIN_ = G_LOG_DOMAIN; > | ^ ~~~~~~~~~~~~ > > whole build log attach. > Oops. Some minor issues were found, and it seems cppgir_clean.sh caused a race condition. I removed it and instead patched the generator of cppgir so that it skips the problematic wrappers; other stuff that can be skipped using cppgir.ignore was moved to files/cppgir_extra_ignore. Thanks kirill@! New tarball attached. With best regards, Nazarenko Mykyta
No comments:
Post a Comment