When will OpenBSD become a friendly place for bug reporters?
Hi! We all know that bugs don't get fixed without backtraces. After few years of using OpenBSD I am annoyed to get mocked for not sending backtraces, but why I don't send them? The answer is: OpenBSD doesn't provide software packages with debugging symbols. Do I look like a Gentoo user? It's not cool to leave no choice to bug reporters but to make them rebuild all ports they use with: $ env CFLAGS='-pipe -g' DEBUG=-g make -j $(sysctl -n hw.ncpu) reinstall The current OpenBSD is definetely not friendly to bug reporters, so don't blame me when I refuse to send backtraces, I am simply not in mood to rebuild software when it shouldn't be necessary, I value my time.
Weird behaviour of updating flavored package via pkg_add -u
Hi! Today is my upgrade day, so I traditionally run `pkg_add -u`, but this time I get a really weird output: Ambiguous: choose package for irssi-1.2.0 a 0: 1: irssi-1.2.1p0 2: irssi-1.2.1p0-socks Your choice: All other packages, including mupdf, upgraded with expeted output: mupdf-1.14.0p2->1.15.0: ok
Re: [Patch] Python for non wxallowed /usr/local
On Tue, May 01, 2018 at 03:13:23AM +0300, IL Ka wrote: > Hi, > > > Can you please give me a link to fix? > Sorry, I misunderstood you. > You said "thank you for fixing py-cryptography". I didn't notice that fix. > I run snapshot and it has wxallowed for /usr/local by default (I haven't > touched it). > $ grep wxallowed /etc/fstab > 4154a363527e316f.h /usr/local ffs rw,wxallowed,nodev 1 2 > > python is wxneeded since 2016 in OpenBSD: > https://github.com/openbsd/ports/blame/master/lang/python/Makefile.inc#L127 > > $ readelf -a $(which python3) | grep WX > OPENBSD_WXNEED 0x 0x 0x > > So, it now works perfectly. > > But virtualenv is broken because it copies file to my home, > and /home is not wxallowed. > > I do not use any package that needs WX, so I am not able to use virtualenv > for no good reason. > > To fix it, I rebuild python from ports using USE_WXNEEDED=no > It solved my problem, but I do not want to build Python from ports, > so better solution would be to > * Accept your patch and create 2 flavors of python > * Or set USE_WXNEEDED=no by default > > Virtualenv is de-facto standard, it is not good to break it > because some modules need WX. > > And enabling "wxneeded" for python generally decreases security. > > > it's better to patch W|X ports > Yes, but is not it better to disable wxneeded and then try to patch them > or create two flavors until they are fixed? > My patch may and will break ports which depend on W|X, so that users have to manually install wx flavor. > Ilya.
Re: [Patch] Python for non wxallowed /usr/local
On Mon, Apr 30, 2018 at 03:22:21PM -0700, Il Ka wrote: > Thank you for fixing py-cryptography in /usr/local, > but we now have python that can't be launched from home > (because it is not wxallowed). > Hi! Can you please give me a link to fix? (I hope you didn't say sarcasm) > And it breaks virtualenv: > https://marc.info/?l=openbsd-misc&m=152510694522191&w=2 > > virtualenv is used by almost any python developer, and we rendered > it to unusable, unless one mount home with wxallowed > or rebuild python from ports disabling this option. > > Even if I do not need py-cryptography I have to pay the price. > > Could we please create two flavors or revert this change? > I can create patch and send it to port maintainer. > This part seems to be complicated, it's better to patch W|X ports to get rid of USE_WXNEEDED in Python. > > > -- > Sent from: > http://openbsd-archive.7691.n7.nabble.com/openbsd-user-ports-f108501.html >
Re: [New] [WIP] Toxcore, Toxic and uTox
On Sun, Apr 29, 2018 at 01:06:21PM +0200, Landry Breuil wrote: > On Sun, Apr 29, 2018 at 12:43:15PM +0200, Landry Breuil wrote: > > On Sun, Apr 29, 2018 at 01:33:57PM +0300, mazocomp wrote: > > > On Sun, Apr 29, 2018 at 01:27:14PM +0300, mazocomp wrote: > > > > On Sun, Apr 29, 2018 at 12:13:16PM +0200, Landry Breuil wrote: > > > > > On Sun, Apr 29, 2018 at 12:41:01PM +0300, mazocomp wrote: > > > > > > On Sat, Apr 28, 2018 at 08:17:56PM -0400, Brian Callahan wrote: > > > > > > > > > > > > > > On 04/28/18 19:43, mazocomp wrote: > > > > > > > > Hi! > > > > > > > > > > > > > > > > Is there any hope this port will be accepted > > > > > > > > or should I keep maintaining it in WIP tree? > > > > > > > > > > > > > > > > I know toxcore is unstable, but I don't know alternatives. > > > > > > > > (except bloated ricochet and retroshare) > > > > > > > > > > > > > > > > > > > > > > toxcore's license marker should be tweaked to GPLv3+, otherwise > > > > > > > ok. > > > > > > > toxic doesn't need HOMEPAGE set if you're using GH_*, otherwise > > > > > > > ok. > > > > > > > utox sets -std=gnu11 so should have COMPILER=base-clang > > > > > > > ports-gcc, also > > > > > > > needs NO_TEST=Yes, otherwise ok. > > > > > > > > > > > > > Ok, done. > > > > > > > > > > > > > I don't know anyone who uses tox so I can't test beyond making > > > > > > > sure they > > > > > > > start up and close without crashing, with both do. > > > > > > > > > > > > > > You are not listed as MAINTAINER of any of these. Do you want to > > > > > > > be > > > > > > > MAINTAINER? > > > > > > > > > > > > > Oh, can I? Ok, I'll try. > > > > > > > > > > I think you need to set USE_LIBTOOL = Yes (or gnu, depending on how it > > > > > uses it) in toxcore so that proper setup of libtool is done. Besides > > > > > that those ports look okay to me. > > > > > > > > > > Landry > > > > > > > > > After testing both of that I can't find file shared_libs.log > > > > (I used find(1)), should I still set USE_LIBTOOL to Yes or gnu? > > > > > > > I'm not sure this is right, but ok: > > > > The rationale is to not explicitely adding libtool from ports (the gnu > > one) unless its really needed, ie delete ports libtool, figure out if > > the port is fine with the base libtool, and then set USE_LIBTOOL=yes > > which will set LIBTOOL in the env during build. If you set it to 'gnu' > > it will also add devel/libtool to the build depends. > > Damn, USE_LIBTOOL=yes is default anyway now. I dont see anything in the port > that requires libtool, and it seems to build fine if gnu libtool is not > installed, so why did you add it to BUILD_DEPENDS ? > I didn't add it to build depends, Dmitriy Czarkoff created this port and added that stuff to build depends, all I did is updating net/toxcore and net/toxic to recent versions and adding recent version of net/utox https://github.com/jasperla/openbsd-wip/commits/master/net/toxcore/Makefile > Landry > > >
Re: [New] [WIP] Toxcore, Toxic and uTox
On Sun, Apr 29, 2018 at 12:43:15PM +0200, Landry Breuil wrote: > On Sun, Apr 29, 2018 at 01:33:57PM +0300, mazocomp wrote: > > On Sun, Apr 29, 2018 at 01:27:14PM +0300, mazocomp wrote: > > > On Sun, Apr 29, 2018 at 12:13:16PM +0200, Landry Breuil wrote: > > > > On Sun, Apr 29, 2018 at 12:41:01PM +0300, mazocomp wrote: > > > > > On Sat, Apr 28, 2018 at 08:17:56PM -0400, Brian Callahan wrote: > > > > > > > > > > > > On 04/28/18 19:43, mazocomp wrote: > > > > > > > Hi! > > > > > > > > > > > > > > Is there any hope this port will be accepted > > > > > > > or should I keep maintaining it in WIP tree? > > > > > > > > > > > > > > I know toxcore is unstable, but I don't know alternatives. > > > > > > > (except bloated ricochet and retroshare) > > > > > > > > > > > > > > > > > > > toxcore's license marker should be tweaked to GPLv3+, otherwise ok. > > > > > > toxic doesn't need HOMEPAGE set if you're using GH_*, otherwise ok. > > > > > > utox sets -std=gnu11 so should have COMPILER=base-clang ports-gcc, > > > > > > also > > > > > > needs NO_TEST=Yes, otherwise ok. > > > > > > > > > > > Ok, done. > > > > > > > > > > > I don't know anyone who uses tox so I can't test beyond making sure > > > > > > they > > > > > > start up and close without crashing, with both do. > > > > > > > > > > > > You are not listed as MAINTAINER of any of these. Do you want to be > > > > > > MAINTAINER? > > > > > > > > > > > Oh, can I? Ok, I'll try. > > > > > > > > I think you need to set USE_LIBTOOL = Yes (or gnu, depending on how it > > > > uses it) in toxcore so that proper setup of libtool is done. Besides > > > > that those ports look okay to me. > > > > > > > > Landry > > > > > > > After testing both of that I can't find file shared_libs.log > > > (I used find(1)), should I still set USE_LIBTOOL to Yes or gnu? > > > > > I'm not sure this is right, but ok: > > The rationale is to not explicitely adding libtool from ports (the gnu > one) unless its really needed, ie delete ports libtool, figure out if > the port is fine with the base libtool, and then set USE_LIBTOOL=yes > which will set LIBTOOL in the env during build. If you set it to 'gnu' > it will also add devel/libtool to the build depends. Oh, I am so blind. Toxcore doesn't use both devel/libtool and devel/check anymore. tox,4.tar.gz Description: application/tar-gz
Re: [New] [WIP] Toxcore, Toxic and uTox
On Sun, Apr 29, 2018 at 01:27:14PM +0300, mazocomp wrote: > On Sun, Apr 29, 2018 at 12:13:16PM +0200, Landry Breuil wrote: > > On Sun, Apr 29, 2018 at 12:41:01PM +0300, mazocomp wrote: > > > On Sat, Apr 28, 2018 at 08:17:56PM -0400, Brian Callahan wrote: > > > > > > > > On 04/28/18 19:43, mazocomp wrote: > > > > > Hi! > > > > > > > > > > Is there any hope this port will be accepted > > > > > or should I keep maintaining it in WIP tree? > > > > > > > > > > I know toxcore is unstable, but I don't know alternatives. > > > > > (except bloated ricochet and retroshare) > > > > > > > > > > > > > toxcore's license marker should be tweaked to GPLv3+, otherwise ok. > > > > toxic doesn't need HOMEPAGE set if you're using GH_*, otherwise ok. > > > > utox sets -std=gnu11 so should have COMPILER=base-clang ports-gcc, also > > > > needs NO_TEST=Yes, otherwise ok. > > > > > > > Ok, done. > > > > > > > I don't know anyone who uses tox so I can't test beyond making sure they > > > > start up and close without crashing, with both do. > > > > > > > > You are not listed as MAINTAINER of any of these. Do you want to be > > > > MAINTAINER? > > > > > > > Oh, can I? Ok, I'll try. > > > > I think you need to set USE_LIBTOOL = Yes (or gnu, depending on how it > > uses it) in toxcore so that proper setup of libtool is done. Besides > > that those ports look okay to me. > > > > Landry > > > After testing both of that I can't find file shared_libs.log > (I used find(1)), should I still set USE_LIBTOOL to Yes or gnu? > I'm not sure this is right, but ok: --- net/toxcore/Makefile.orig Sun Apr 29 13:18:00 2018 +++ net/toxcore/MakefileSun Apr 29 13:31:51 2018 @@ -31,6 +31,8 @@ LIB_DEPENDS = audio/opus \ multimedia/libvpx \ security/libsodium +USE_LIBTOOL = Yes + NO_TEST = Yes .include
Re: [New] [WIP] Toxcore, Toxic and uTox
On Sun, Apr 29, 2018 at 12:13:16PM +0200, Landry Breuil wrote: > On Sun, Apr 29, 2018 at 12:41:01PM +0300, mazocomp wrote: > > On Sat, Apr 28, 2018 at 08:17:56PM -0400, Brian Callahan wrote: > > > > > > On 04/28/18 19:43, mazocomp wrote: > > > > Hi! > > > > > > > > Is there any hope this port will be accepted > > > > or should I keep maintaining it in WIP tree? > > > > > > > > I know toxcore is unstable, but I don't know alternatives. > > > > (except bloated ricochet and retroshare) > > > > > > > > > > toxcore's license marker should be tweaked to GPLv3+, otherwise ok. > > > toxic doesn't need HOMEPAGE set if you're using GH_*, otherwise ok. > > > utox sets -std=gnu11 so should have COMPILER=base-clang ports-gcc, also > > > needs NO_TEST=Yes, otherwise ok. > > > > > Ok, done. > > > > > I don't know anyone who uses tox so I can't test beyond making sure they > > > start up and close without crashing, with both do. > > > > > > You are not listed as MAINTAINER of any of these. Do you want to be > > > MAINTAINER? > > > > > Oh, can I? Ok, I'll try. > > I think you need to set USE_LIBTOOL = Yes (or gnu, depending on how it > uses it) in toxcore so that proper setup of libtool is done. Besides > that those ports look okay to me. > > Landry > After testing both of that I can't find file shared_libs.log (I used find(1)), should I still set USE_LIBTOOL to Yes or gnu?
Re: [New] [WIP] Toxcore, Toxic and uTox
The only question left is: who will commit this? I don't have write access.
Re: [New] [WIP] Toxcore, Toxic and uTox
On Sat, Apr 28, 2018 at 08:17:56PM -0400, Brian Callahan wrote: > > On 04/28/18 19:43, mazocomp wrote: > > Hi! > > > > Is there any hope this port will be accepted > > or should I keep maintaining it in WIP tree? > > > > I know toxcore is unstable, but I don't know alternatives. > > (except bloated ricochet and retroshare) > > > > toxcore's license marker should be tweaked to GPLv3+, otherwise ok. > toxic doesn't need HOMEPAGE set if you're using GH_*, otherwise ok. > utox sets -std=gnu11 so should have COMPILER=base-clang ports-gcc, also > needs NO_TEST=Yes, otherwise ok. > Ok, done. > I don't know anyone who uses tox so I can't test beyond making sure they > start up and close without crashing, with both do. > > You are not listed as MAINTAINER of any of these. Do you want to be > MAINTAINER? > Oh, can I? Ok, I'll try. > ~Brian > tox,3.tar.gz Description: application/tar-gz
Re: [New] [WIP] Toxcore, Toxic and uTox
Hi! Is there any hope this port will be accepted or should I keep maintaining it in WIP tree? I know toxcore is unstable, but I don't know alternatives. (except bloated ricochet and retroshare)
Re: [New] www/vimb 3.1.0
On Wed, Apr 25, 2018 at 09:56:20PM +0200, Erling Westenvik wrote: > On Tue, Apr 24, 2018 at 02:44:51PM +0300, mazocomp wrote: > > I'll try once more to submit this port (this is a port at WIP's pull > > request, it is ignored). > > Vimb is modal webkit-based browser inspired by vim. > > Lovely. Thank you! Compiled and installed without problems (except for > taking ages). Vimperator is doomed and only lives on short borrowed > time requiring a hopelessly slow and bloated Firefox ESR, and I can't > stand chrome and vrome. > > Might a pkg-readme describing plugins needed for video, be something to > consider? Afaik: > > gstreamer1-plugins-good for Youtube videos. > gstreamer1-plugins-libav for Facebook videos. > > Erling > Don't know, but this is specific to webkit engine, so I think someone should add pkg-readme to webkit ports, but isn't just saying Youtube and Facebook videos unfair? About bloated web browsers, webkit itself is not lightweight and comes from the most bloated software called KDE :( Anyway, Lynx is still here <3
Re: [New] [WIP] Toxcore, Toxic and uTox
Let me know if there is anything else I miss, I hope these ports need not stay at WIP tree.
Re: [New] [WIP] Toxcore, Toxic and uTox
On Wed, Apr 25, 2018 at 01:31:54PM +0100, Stuart Henderson wrote:
> On 2018/04/25 15:26, mazocomp wrote:
> > > >=< toxic:
> > >
> > > ignores CFLAGS
> > >
> > I am stuck here. Do you mean CFLAGS is not setable externally?
>
> The port should be using flags coming from CFLAGS in the enivronment,
> by default this is -O2 -pipe, but can be overridden with e.g. make
> CFLAGS="-O1"
>
> It looks like in this case it needs
>
> MAKE_ENV += USER_CFLAGS="${CFLAGS}"
>
>
Thank you, here I go:
tox,2.tar.gz
Description: application/tar-gz
Re: [New] [WIP] Toxcore, Toxic and uTox
On Wed, Apr 25, 2018 at 12:12:10PM +0100, Stuart Henderson wrote:
> On 2018/04/25 13:15, mazocomp wrote:
> > On Wed, Apr 25, 2018 at 10:56:07AM +0100, Stuart Henderson wrote:
> > > On 2018/04/25 12:51, mazocomp wrote:
> > > > > - net/toxic: use the python module, don't RUN_DEPENDS +=
> > > > > lang/python/3.6
> > > > >
> > > > It's an error to use python module in this port:
> > >
> > > It's an error to _not_ use the python module in a port that uses python
> > > because it will cause pain when we switch versions - easy fix for this
> > > though:
> > >
> > > > ===> Building for toxic-0.8.2
> > > > /usr/local/bin/python3.6: can't open file './setup.py': [Errno 2] No
> > > > such file or directory
> > > > *** Error 2 in . (/usr/ports/lang/python/python.port.mk:185 'do-build':
> > > > @: ;mkdir -p /tmp/pobj/toxic-0.8.2-python3/lib/python3.6/site-packag...)
> > > > *** Error 1 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2736
> > > > '/tmp/pobj/toxic-0.8.2-python3/.build_done')
> > > > *** Error 1 in /usr/ports/openbsd-wip/net/toxic
> > > > (/usr/ports/infrastructure/mk/bsd.port.mk:2417 'all')
> > >
> > > set CONFIGURE_STYLE=none.
> > >
> >
> > Ok, I am done. Also I had a pain with MODPY_VERSION variable, so I just
> > renamed flavor python to python3.
> >
> > Here is corrected version of these ports:
>
> Thanks, so now I've done a build test and noticed a couple other things
>
> >=< toxcore:
>
> "SEPATATE_BUILD" - typo, plus it is set by default for cmake ports, please
> remove
> the line.
>
Ok.
> TEST_TARGET = check
> doesn't work; please either repair or use NO_TEST=Yes instead
>
> there are some time_t problems but only in the auto_tests dir.
> not a showstopper, but even if tests are fixed they'll probably
> fail on ILP32 arches.
>
> COMPILER_LIBCXX seems unneeded; only used in (not-installed) test files?
>
I didn't know what extra word means, all I did was updating the work of
Dmitriy Czarkoff and Pascal Stumpf. Ok, I set NO_TEST to Yes
> >=< toxic:
>
> ignores CFLAGS
>
I am stuck here. Do you mean CFLAGS is not setable externally?
> python3 flavour is missing LIB_DEPENDS on ${MODPY_LIB_DEPENDS} (masked by
> RUN_DEPENDS), also missing WANTLIB on ${MODPY_WANTLIB} intl util
>
Ok.
> FLAVOR=python3 or FLAVOR="python3 no_x11" just produces a package named
> toxic-0.8.2 or toxic-0.8.2-no_x11 (check with make show=PKGNAMES with
> FLAVOR set in various ways). This is a bad interaction with python.port.mk's
> python3 package support. One possible fix would be to name the flavour "py"
> instead of python or python3.
>
> FLAVOR=no_x11 has Extra: vpx
>
> FLAVOR="notify no_x11" has Extra: m vpx
>
> There doesn't seem a lot of point in no_x11 if it's going to pull in
> desktop-file-utils; is there really an advantage to having all these
> flavours anyway? they do make testing and updates more complicated.
>
Having notify flavor has only advantage not to install libnotify, no_x11
has minor advantage: if Toxic runs in tmux session, it may crash if
you'll stop running terminal emulator at X11 session. Ok, I am removing
all these flavors.
> >=< uTox:
>
> overrides optimizer flags from CFLAGS by adding -Os, otherwise looks good
>
>
Ok.
Re: [New] [WIP] Toxcore, Toxic and uTox
Oups, one more thing I forgot:
--- net/utox/Makefile.orig Wed Apr 25 13:47:06 2018
+++ net/utox/Makefile Wed Apr 25 13:47:17 2018
@@ -5,7 +5,7 @@ COMMENT = lightweight Tox client
V =0.17.0
DISTNAME = uTox-${V}-full
-PKGNAME = uTox-${V}
+PKGNAME = utox-${V}
MASTER_SITES = https://github.com/uTox/uTox/releases/download/v${V}/
CATEGORIES = net
Re: [New] [WIP] Toxcore, Toxic and uTox
On Wed, Apr 25, 2018 at 10:56:07AM +0100, Stuart Henderson wrote: > On 2018/04/25 12:51, mazocomp wrote: > > > - net/toxic: use the python module, don't RUN_DEPENDS += lang/python/3.6 > > > > > It's an error to use python module in this port: > > It's an error to _not_ use the python module in a port that uses python > because it will cause pain when we switch versions - easy fix for this > though: > > > ===> Building for toxic-0.8.2 > > /usr/local/bin/python3.6: can't open file './setup.py': [Errno 2] No > > such file or directory > > *** Error 2 in . (/usr/ports/lang/python/python.port.mk:185 'do-build': > > @: ;mkdir -p /tmp/pobj/toxic-0.8.2-python3/lib/python3.6/site-packag...) > > *** Error 1 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2736 > > '/tmp/pobj/toxic-0.8.2-python3/.build_done') > > *** Error 1 in /usr/ports/openbsd-wip/net/toxic > > (/usr/ports/infrastructure/mk/bsd.port.mk:2417 'all') > > set CONFIGURE_STYLE=none. > Ok, I am done. Also I had a pain with MODPY_VERSION variable, so I just renamed flavor python to python3. Here is corrected version of these ports: tox,1.tar.gz Description: application/tar-gz
Re: [New] [WIP] Toxcore, Toxic and uTox
On Wed, Apr 25, 2018 at 09:18:18AM +0100, Stuart Henderson wrote: > On 2018/04/25 08:28, mazocomp wrote: > > Hi! > > > > This time I'll submit three ports because uTox's upstream has finally > > uploaded their own tarball at github (auto-generated one didn't include > > submodules, so failed to compile, I can't work with multiple DISTFILES). > > > > Tox is secure distributed messaging protocol with audio and video chat > > capabilities. > > > > The attached ports are: > > > > * net/toxcore Tox client library. > > * net/toxicncurses-based Tox client. > > * net/utox Lightweight Tox client. > > > > There are caveats worth to mention: > > > > * Thanks to audio/openal, audio input won't work at Toxic and uTox (how > >old is this issue?). > > * Don't expect push-to-talk work at uTox, it has Linux-specific code. > > * uTox has GTK+3 as optional dependency (to make filepicker work). > > > > Anyway, these ports are ready and wait for import. > > They're not ready yet. > > - net/toxcore: start SHARED_LIBS at 0.0 > Ok. > - net/toxic, net/toxore: put GH_* near the top of the file, alongside > PKGNAME/DISTNAME where those are also used, or in the same place they > would appear otherwise > Ok. > - net/toxic: use the python module, don't RUN_DEPENDS += lang/python/3.6 > It's an error to use python module in this port: ===> Building for toxic-0.8.2 /usr/local/bin/python3.6: can't open file './setup.py': [Errno 2] No such file or directory *** Error 2 in . (/usr/ports/lang/python/python.port.mk:185 'do-build': @: ;mkdir -p /tmp/pobj/toxic-0.8.2-python3/lib/python3.6/site-packag...) *** Error 1 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2736 '/tmp/pobj/toxic-0.8.2-python3/.build_done') *** Error 1 in /usr/ports/openbsd-wip/net/toxic (/usr/ports/infrastructure/mk/bsd.port.mk:2417 'all') > - net/utox: add blank line after RCS id, lowercase start of COMMENT, > zap stray \ in LIB_DEPENDS > Ok.
[New] [WIP] Toxcore, Toxic and uTox
Hi! This time I'll submit three ports because uTox's upstream has finally uploaded their own tarball at github (auto-generated one didn't include submodules, so failed to compile, I can't work with multiple DISTFILES). Tox is secure distributed messaging protocol with audio and video chat capabilities. The attached ports are: * net/toxcore Tox client library. * net/toxicncurses-based Tox client. * net/utox Lightweight Tox client. There are caveats worth to mention: * Thanks to audio/openal, audio input won't work at Toxic and uTox (how old is this issue?). * Don't expect push-to-talk work at uTox, it has Linux-specific code. * uTox has GTK+3 as optional dependency (to make filepicker work). Anyway, these ports are ready and wait for import. tox.tar.gz Description: application/tar-gz
Re: [New] www/vimb 3.1.0
That PR is no longer ignored, it's merged.
[New] www/vimb 3.1.0
Hi! I'll try once more to submit this port (this is a port at WIP's pull request, it is ignored). Vimb is modal webkit-based browser inspired by vim. vimb.tar.gz Description: application/tar-gz
[New] net/toxcore 0.2.2 and net/toxic 0.8.2
Hi! net/toxcore: Tox client library Tox is secure distributed messaging protocol with audio and video chat capabilities. net/toxic: ncurses-based Tox client Toxic is a Tox-based instant messenging client which formerly resided in the Tox core repository, and is now available as a standalone application. Flavors: python - build with Python scripting support no_x11 - build without X11 support notify - build with dekstop notifications support tox.tar.gz Description: application/tar-gz
Re: [rc] emacs-26.1-rc1
Emacs still doesn't support sound? (I've noticed --without-sound option)
Re: games/minetest: questions about patch
Oh, ok, thanks. I've sent patch for src/unittest/test_serialization.cpp earlier, waiting for merge.
Re: games/minetest: questions about patch
@@ -682,14 +689,12 @@ else()
set(OTHER_FLAGS "-mthreads -fexceptions")
endif()
- set(CMAKE_CXX_FLAGS_RELEASE "-DNDEBUG ${RELEASE_WARNING_FLAGS}
${WARNING_FLAGS} ${OTHER_FLAGS} -ffast-math -Wall -pipe -funroll-loops")
+ set(CMAKE_CXX_FLAGS_RELEASE "-DNDEBUG ${RELEASE_WARNING_FLAGS}
${WARNING_FLAGS} ${OTHER_FLAGS}")
Here is removed -Wall flag (in case your mail client doesn't display
lines longer than 72 characters, here is formatted version).
I've sent other patches to Minetest's developers. I think this port
will be updated when Minetest 0.4.17 will be released (I tried to port
0.4.16, while I don't know C++ I can't fix 0.4.16, I catch abort trap
when trying enter the world).
games/minetest: questions about patch
I had to patch it to make it compile and run.*
games/minetest: questions about patch
Hi!
I am running Minetest from branch backport-0.4, I had to patch it make
it compile and run. Since I am ignorant I have questions about patch
for src/CMakeLists.txt (I've just compiled this game without patching
this file and game runs fine)
@@ -61,6 +61,8 @@ set(USE_GETTEXT FALSE)
if(ENABLE_GETTEXT)
find_package(GettextLib)
+ find_package(Iconv)
+ find_package(Intl)
if(GETTEXT_FOUND)
if(WIN32)
message(STATUS "GetText library: ${GETTEXT_LIBRARY}")
These two lines are not required, why add them?
@@ -550,7 +555,8 @@ if(BUILD_CLIENT)
${BZIP2_LIBRARIES}
${PNG_LIBRARIES}
${X11_LIBRARIES}
- ${GETTEXT_LIBRARY}
+ ${ICONV_LIBRARIES}
+ ${Intl_LIBRARIES}
${SOUND_LIBRARIES}
${SQLITE3_LIBRARY}
${LUA_LIBRARY}
@@ -612,7 +618,8 @@ if(BUILD_SERVER)
${ZLIB_LIBRARIES}
${SQLITE3_LIBRARY}
${JSON_LIBRARY}
- ${GETTEXT_LIBRARY}
+ ${ICONV_LIBRARIES}
+ ${Intl_LIBRARIES}
${LUA_LIBRARY}
${GMP_LIBRARY}
${PLATFORM_LIBS}
Why remove ${GETTEXT_LIBRARY} lines? Also is that patch worth for
if(BUILD_SERVER) if server is not actually built (I checked plist)?
@@ -682,14 +689,12 @@ else()
set(OTHER_FLAGS "-mthreads -fexceptions")
endif()
- set(CMAKE_CXX_FLAGS_RELEASE "-DNDEBUG ${RELEASE_WARNING_FLAGS}
${WARNING_FLAGS} ${OTHER_FLAGS} -ffast-math -Wall -pipe -funroll-loops")
+ set(CMAKE_CXX_FLAGS_RELEASE "-DNDEBUG ${RELEASE_WARNING_FLAGS}
${WARNING_FLAGS} ${OTHER_FLAGS}")
if(APPLE)
set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -Os")
- else()
- set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -O3
-fomit-frame-pointer")
endif(APPLE)
set(CMAKE_CXX_FLAGS_SEMIDEBUG "-g -O1 -Wall -Wabi ${WARNING_FLAGS}
${OTHER_FLAGS}")
- set(CMAKE_CXX_FLAGS_DEBUG "-g -O0 -Wall -Wabi ${WARNING_FLAGS}
${OTHER_FLAGS}")
+ set(CMAKE_CXX_FLAGS_DEBUG "-g -Wall -Wabi ${WARNING_FLAGS}
${OTHER_FLAGS}")
if(USE_GPROF)
set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -pg")
Is that -Wall flag removed to make compilation faster?
What's wrong about -O0 coming after -g? And why not remove -O1 for
semi-debug in that case?
And what's wrong with those optimisations?
Let me guess, this game doesn't compile/run because of old GCC on some
architectures, does it?
games/supertux is broken
mazocomp$ supertux2 [WARNING] /usr/obj/ports/supertux-0.5.1/SuperTux-v0.5.1-Source/src/supertux/main.cpp:188 Couldn't add './share/supertux2' to physfs searchpath: not found [FATAL] /usr/obj/ports/supertux-0.5.1/SuperTux-v0.5.1-Source/src/supertux/main.cpp:510 Unexpected exception: Couldn't open 'images/engine/icons/supertux-256x256.png': not found mazocomp$
Re: mplayer is gone?
On Tue, Apr 10, 2018 at 09:54:21PM +0300, Dumitru Mișu Moldovan wrote: > On 04/10/18 19:59, mazocomp wrote: > > On Tue, Apr 10, 2018 at 06:58:37PM +0300, Dumitru Mișu Moldovan wrote: > > […] > > >> Is mplayer still a thing? Have used it for many years a long time ago, > >> but now I see the latest stable is 2 years old and nicknamed > >> "worksforme"… :-] > >> > >> I think of mpv as the spiritual successor of mplayer, being the most > >> successful fork. Is there anything mplayer does and mpv cannot? Using > >> something else (ffmpeg's ffplay) only when in need of best performance > >> possible on my old hardware here. > >> > >> > > > > Yes, it is. > > > > MPlayer has less dependencies and also I tested both mpv and mplayer at > > hardware which didn't have 2D acceleration. That was terrible, mpv > > tried to put every frame at screen, so giving 1 minute video to mpv made > > video last 6 minutes, but it could play all audio in 1 minute, so I > > didn't hear any sound last 5 minutes. I didn't notice the same problem > > at mplayer, it played as much frames as possible in 1 minute. Maybe I > > should read mpv's man page, but I like when everything works by default. > > > > OK, I can relate to that… Give ffplay a try though, it gives me more > performance compared to mplayer/mpv and has even less dependencies (just > install ffmpeg). It's a very basic player, but it works. > > Although I'm not sure what you mean by no 2D acceleration, maybe "-vo > x11" in mplayer/mpv parlance? Everything I use these days has at least > XVideo support. Only had to use "-vo x11" when the Linux kernel video > driver was crashing my desktop when waking from resume lately and I > wasn't ready for a reboot to make video acceleration work again. > > Got so used with it that I thought it was a hardware issue. Amazed to > see no problems with waking up in OpenBSD… Kudos to whoever made that > possible, OpenBSD's ACPI support is alien tech compared to what else is > available in the BSDs, and for me it even beats Linux's implementation. > Oh, thank you. ffplay has basically everything I need, except controlling sound volume with keys, but that's fine. About 2D acceleration, radeondrm(4) doesn't support radeons with GCN architecture: it can't provide 2D and 3D acceleration, so even playing video is tough to computer (I tried to play supertux, I had to make my window tiny to have playable performance, but still overheat).
Re: mplayer is gone?
On Tue, Apr 10, 2018 at 06:58:37PM +0300, Dumitru Mișu Moldovan wrote: > On 04/10/18 17:51, mazoc...@disroot.org wrote: > > Hi! > > > > Is this happening because of Clang 6 fallout? > > > > mazocomp$ pkg_add -n mplayer > > quirks-2.418 signed on 2018-04-09T22:17:28Z > > Can't find mplayer > > mazocomp$ > > > > Is mplayer still a thing? Have used it for many years a long time ago, > but now I see the latest stable is 2 years old and nicknamed > "worksforme"… :-] > > I think of mpv as the spiritual successor of mplayer, being the most > successful fork. Is there anything mplayer does and mpv cannot? Using > something else (ffmpeg's ffplay) only when in need of best performance > possible on my old hardware here. > > Yes, it is. MPlayer has less dependencies and also I tested both mpv and mplayer at hardware which didn't have 2D acceleration. That was terrible, mpv tried to put every frame at screen, so giving 1 minute video to mpv made video last 6 minutes, but it could play all audio in 1 minute, so I didn't hear any sound last 5 minutes. I didn't notice the same problem at mplayer, it played as much frames as possible in 1 minute. Maybe I should read mpv's man page, but I like when everything works by default.
mplayer is gone?
Hi! Is this happening because of Clang 6 fallout? mazocomp$ pkg_add -n mplayer quirks-2.418 signed on 2018-04-09T22:17:28Z Can't find mplayer mazocomp$
MPlayer is gone?
Hi! Is this happening because of Clang 6 fallout? mazocomp$ pkg_add -n mplayer quirks-2.418 signed on 2018-04-09T22:17:28Z Can't find mplayer mazocomp$ uname -a OpenBSD mazocomp.pc 6.3 GENERIC.MP#159 amd64 mazocomp$
[Patch] games/urbanterror doesn't rely on W|X
Hi! By quickly grepping Urban Terror's code I've seen: ./code/qcommon/vm_x86.c:if(mprotect(vm->codeBase, compiledOfs, PROT_READ|PROT_EXEC)) ./code/qcommon/vm_x86_64.c: mem = mmap(NULL, stb.st_size, PROT_READ|PROT_EXEC, MAP_SHARED, fd, 0); ./code/qcommon/vm_x86_64.c: if(mprotect(vm->codeBase, compiledOfs, PROT_READ|PROT_EXEC)) After that I removed USE_WXNEEDED from port's Makefile and tested this game (my /usr/local is not mounted under wxallowed). I enjoyed playing this game, no W^X violations noticed. Index: Makefile === RCS file: /cvs/ports/games/urbanterror/Makefile,v retrieving revision 1.4 diff -u -p -r1.4 Makefile --- Makefile8 Feb 2017 12:55:18 - 1.4 +++ Makefile2 Feb 2018 04:26:18 - @@ -1,6 +1,5 @@ # $OpenBSD: Makefile,v 1.4 2017/02/08 12:55:18 sthen Exp $ -USE_WXNEEDED = Yes ONLY_FOR_ARCHS = amd64 i386 COMMENT = fast-paced hollywood-tactical shooter: "fun over realism"
Re: [New] editors/emacs-lite
On Tue, Jan 30, 2018 at 07:39:37PM +0100, Matthieu Herrb wrote: > On Tue, Jan 30, 2018 at 02:40:06PM +0200, mazocomp wrote: > > On Mon, Jan 29, 2018 at 10:48:41PM -0500, Brian Callahan wrote: > > > I'm breaking my own rule regarding emails like this by even responding, > > > but...$ man 1 mg > > > > I know about mg(1), but you know that emacs is not only the text editor. > > > > Emacs has extensions that make it usable, for example, as IRC client. > > > > Also mg(1) doesn't support UTF-8 and I write russian characters > > somehow (well, I find editors/jed good for this). > > What's wrong with emacs-21.4p37-no_x11 ? it has no extra run > dependencies AFAICT ? > -- > Matthieu Herrb I allready told that both Emacs 21 and XEmacs 21 don't print russian characters and setting "prefer-coding-system" doesn't help.
Re: [New] editors/emacs-lite
On Tue, Jan 30, 2018 at 10:29:47AM -0300, Gonzalo L. Rodriguez wrote: > On [30/01/18] [12:40P], mazocomp wrote: > ; On Mon, Jan 29, 2018 at 10:48:41PM -0500, Brian Callahan wrote: > ; > I'm breaking my own rule regarding emails like this by even responding, > but...$ man 1 mg > ; > ; I know about mg(1), but you know that emacs is not only the text editor. > ; > ; Emacs has extensions that make it usable, for example, as IRC client. > ; > ; Also mg(1) doesn't support UTF-8 and I write russian characters > ; somehow (well, I find editors/jed good for this). > ; > > editors/zile? > > -- > Sending from my toaster. Zile doesn't support UTF-8 either. I also tried both Emacs 21 and XEmacs 21, they are as light as this emacs-lite, but trying to write russian characters printed me squares. (of course I tried setting prefer-coding-system) editors/uemacs annoyingly separates UTF-8 characters by space.
Re: [New] editors/emacs-lite
On Mon, Jan 29, 2018 at 10:48:41PM -0500, Brian Callahan wrote: > I'm breaking my own rule regarding emails like this by even responding, > but...$ man 1 mg I know about mg(1), but you know that emacs is not only the text editor. Emacs has extensions that make it usable, for example, as IRC client. Also mg(1) doesn't support UTF-8 and I write russian characters somehow (well, I find editors/jed good for this).
[New] editors/emacs-lite
Hi! Emacs has too many dependencies which make it as heavy as cow or even elephant... I was thinking about adding a flavor, but it seems to be a headache. So, I've decided to make a separate package (by copying editors/emacs) to make Emacs as light as pig... emacs-lite.tar.gz Description: application/tar-gz
Re: [Patch] Python for non wxallowed /usr/local
Landry, if I install unpatched port, it refuses to run with "permission denied" error. In this case I have to forget about binary updating and compile this port manually.
Re: [Patch] Python for non wxallowed /usr/local
On Fri, Jan 26, 2018 at 02:56:11PM +, Stuart Henderson wrote: > On 2018/01/26 16:36, mazocomp wrote: > > +Flavor wx - build with W|X reliance. > > Of course we have thought about this before! But it's not as simple as > you propose. Done in the way you suggest as a simple either/or flavour > is going to cause too many problems and confusion with dependent ports > and isn't acceptable. > > Maybe we could consider something that produces two binaries, one with > wxneeded as "/usr/local/bin/pythonX.Y-wx" that we can use for things that > need webkit modules etc, and one without wxneeded as the standard > "pythonX.Y". > Hm, every port which relies on W|X will be patched to use python*-wx binary, won't it? Sounds great.
Re: [Patch] Python for non wxallowed /usr/local
On Fri, Jan 26, 2018 at 03:49:48PM +0100, Solène Rapenne wrote:
> Le 2018-01-26 15:36, mazocomp a écrit :
> > # Python itself is clean, but some extensions e.g. py-cryptography
> > # and QtWebKit require W|X mappings.
> > +.if ${FLAVOR} == "wx"
> > USE_WXNEEDED = Yes
> > +.endif
> >
> > .if ${VERSION} == "3.6"
> > ALL_TARGET = all
>
> IIRC python doesn't have W^X issue but some python libraries will load
> shared libraries requiring W^X so python has to be compiled with wxneeded
> to load those libraries.
The comment in Makefile says that clearly. But I prefer using Python
without wxallowed /usr/local, I run software which doesn't rely on W|X.
[Patch] Python for non wxallowed /usr/local
diff --git lang/python/2.7/pkg/DESCR-main lang/python/2.7/pkg/DESCR-main
index adfd63f9840..9d7c54a9e93 100644
--- lang/python/2.7/pkg/DESCR-main
+++ lang/python/2.7/pkg/DESCR-main
@@ -12,3 +12,5 @@ written in C or C++. On most systems such modules may be
dynamically loaded. Python is also adaptable as an
extension language for existing applications. See the
internal documentation for hints.
+
+Flavor wx - build with W|X reliance.
diff --git lang/python/3.6/pkg/DESCR-main lang/python/3.6/pkg/DESCR-main
index adfd63f9840..9d7c54a9e93 100644
--- lang/python/3.6/pkg/DESCR-main
+++ lang/python/3.6/pkg/DESCR-main
@@ -12,3 +12,5 @@ written in C or C++. On most systems such modules may be
dynamically loaded. Python is also adaptable as an
extension language for existing applications. See the
internal documentation for hints.
+
+Flavor wx - build with W|X reliance.
diff --git lang/python/Makefile.inc lang/python/Makefile.inc
index f66fda7e47f..7f317e4b6e2 100644
--- lang/python/Makefile.inc
+++ lang/python/Makefile.inc
@@ -43,6 +43,9 @@ MAINTAINER = Remi Pointel
# A summary of changes to Python is required to be distributed
PERMIT_PACKAGE_CDROM = Yes
+FLAVORS= wx
+FLAVOR?= wx
+
.if ${VERSION} == "2.7"
LIB_SUFX =
MULTI_PACKAGES = -main -tests -tools -gdbm -idle -tkinter -bsddb
@@ -124,7 +127,9 @@ FAKE_FLAGS += RANLIB=:
# Python itself is clean, but some extensions e.g. py-cryptography
# and QtWebKit require W|X mappings.
+.if ${FLAVOR} == "wx"
USE_WXNEEDED = Yes
+.endif
.if ${VERSION} == "3.6"
ALL_TARGET = all
NEW: www/vimb
Hi! Vimb is modal webkit-based browser inspired by vim This is latest 3.0-alpha release which depends on www/webkitgtk4 vimb.tar.gz Description: application/tar-gz
NEW: Tox
Hi! Tox is secure distributed messaging protocol with audio and video chat capabilities. The attached ports are: * net/toxcore Tox client library * net/toxicAn ncurses-based Tox client I have to warn you: don't compile the "sound" flavor, it depends on openal which causes segmentation faults, but 1.16.0 version of openal port doesn't cause segmentation faults, also openal doesn't support sndio backend, so you won't be able to speak, but you'll be able to listen. tox.tar.gz Description: application/tar-gz
QtWebEngine doesn't build
Hi! Why do you keep x11/qt5/qtwebengine in ports tree if ti doesn't build?
