On Sun, Jan 01, 2017 at 06:18:15PM +0100, Landry Breuil wrote:
> On Sat, Dec 31, 2016 at 12:45:52PM +0100, Landry Breuil wrote:
> > On Sat, Dec 31, 2016 at 04:38:10PM +0500, Alexandr Shadchin wrote:
> > > On Sat, Dec 31, 2016 at 11:43:00AM +0100, Landry Breuil wrote:
> > > > On Thu, Dec 29, 2016 at 07:19:38PM +0500, Alexandr Shadchin wrote:
> > > > > On Thu, Dec 29, 2016 at 03:01:12PM +0100, Landry Breuil wrote:
> > > > > > On Thu, Dec 29, 2016 at 02:53:04PM +0100, Remi Pointel wrote:
> > > > > > > On 12/29/16 14:00, Alexandr Shadchin wrote:
> > > > > > > > On Mon, Dec 26, 2016 at 11:33:43PM +0500, Alexandr Shadchin 
> > > > > > > > wrote:
> > > > > > > > > On Mon, Dec 26, 2016 at 07:18:54PM +0100, Remi Pointel wrote:
> > > > > > > > > > On 12/26/16 19:12, Alexandr Shadchin wrote:
> > > > > > > > > > > Hi,
> > > > > > > > > > > 
> > > > > > > > > > > What is the timeline of flipping py3 flavour
> > > > > > > > > > > over to python3.5?
> > > > > > > > > > > 
> > > > > > > > > > 
> > > > > > > > > > Hi,
> > > > > > > > > > 
> > > > > > > > > > I would like to switch the default python 3 version to 3.6 
> > > > > > > > > > (I think 3.6 will
> > > > > > > > > > be imported these days).
> > > > > > > > > > 
> > > > > > > > > > We should import python 3.6, then run a bulk build and fix 
> > > > > > > > > > the ports don't
> > > > > > > > > > build. Then we will remove python 3.4 and 3.5 if we can.
> > > > > > > > > > 
> > > > > > > > > > Remi.
> > > > > > > > > 
> > > > > > > > > For me, it's even better :-)
> > > > > > > > > 
> > > > > > > > > I have two weeks free, I can work with fails.
> > > > > > > > > 
> > > > > > > > 
> > > > > > > > I tried switch on Python 3.6, for me works fine.
> > > > > > > > 
> > > > > > > > 1. Fix pyo-files (1_fix_pyo.diff)
> > > > > > > >    For more details read PEP 488 [1]
> > > > > > > > 2. Fix py3-cairo (2_fix_py3-cairo.diff)
> > > > > > > > 3. Fix py-tornado (3_fix_tornado.diff)
> > > > > > > > 4. Switch on Python 3.6
> > > > > > > > 
> > > > > > > > [1] - https://www.python.org/dev/peps/pep-0488/
> > > > > > > > 
> > > > > > > 
> > > > > > > Sounds good for me, this diff should be tested by landry@ in a 
> > > > > > > bulk.
> > > > > > > 
> > > > > > > Thank you for your work.
> > > > > > 
> > > > > > Does this diff has *all* the REVISION bumps needed ?
> > > > > > I'll put it in the next bulk..
> > > > > > 
> > > > > > Landry
> > > > > 
> > > > > Yes, on clean system.
> > > > > 
> > > > > For my laptop was required bump py-setuptools.
> > > > > But I'm a noob at this, so it may have missed something :-)
> > > > 
> > > > So im about to start a bulk with this, but some things that worry me:
> > > > 
> > > > - you *really* should explain your diffs. the py-tornado looks like
> > > >   magic, why suddenly would the egg-info/requires.txt file be not
> > > > installed anymore ? wont it affect other ports ?
> > > 
> > > py-tornado depend:
> > > py3.4 need py-backport-abc (requires.txt generated)
> > > py3.6 not need anything (requires.txt not generated)
> > 
> > ok, thant makes sense :)
> > 
> > > > - the py3-cairo diff looks like magic. i know waf sucks, but why is this
> > > >   return needed ?
> > > 
> > > this patch from FreeBSD, that was obtained from Ubuntu repository :)
> > > 
> > > https://svnweb.freebsd.org/ports/head/graphics/py3-cairo/files/pycairo-1.10.0-waf-py3_5.patch?view=log
> > 
> > great cargo culting :)
> > 
> > > > - you didnt bump *any* of the other ports depending on py3, so they will
> > > >   all break because of the plist db. This sucks, because i'll have to do
> > > > the work (as i explained to remi).
> > > 
> > > I can make diff for bump all py3 ports.
> > > 
> > > > - i suppose you didnt build any of the other ports depending on py3
> > > >   besides setuptools ?
> > > 
> > > I built this py3 ports:
> > > sqlite3 /usr/local/share/sqlports "select distinct fullpkgpath from 
> > > depends where dependspath like '%python/3.4%'"
> > 
> > great, but had you built them *before* with py34 so that they were all
> > registered in your plistdb ? that's the whole point ....
> > 
> > anyway, starting the bulk with a portstree updated to
> > https://github.com/openbsd/ports/commit/6864fa0e1c5e96a8095dda077af04c0b30a6ccc2
> > 
> > with this diff on top:
> > http://rhaalovely.net/~landry/stuff/py36.fulldiff
> > 
> > curated list passed to portbump from sqlports, and removed all the
> > flavors/subpackages.
> 
> Fwiw, after 25h and 4300 pkgs, the only 'important' fallout is
> libreoffice, something about pyuno:
> 
> checking whether /usr/local/bin/python3.6 version is >= 2.6... yes
> checking for /usr/local/bin/python3.6 version... 3.6
> checking for /usr/local/bin/python3.6 platform... openbsd6
> checking for /usr/local/bin/python3.6 script directory... 
> ${prefix}/lib/python3.6/site-packages
> checking for /usr/local/bin/python3.6 extension module directory... 
> ${exec_prefix}/lib/python3.6/site-packages
> checking which Python to use for Pyuno... checking for a Python interpreter 
> with version >= 3.3... none
> internal
> 
> ...
> but later on it tries to build a local python 3.5 ?
> ...
> mkdir -p 
> /usr/obj/ports/libreoffice-5.2.4.2/libreoffice-5.2.4.2/workdir/ExternalProject/mythes/
> touch 
> /usr/obj/ports/libreoffice-5.2.4.2/libreoffice-5.2.4.2/workdir/ExternalProject/python3.prepare
> gmake[1]: *** No rule to make target 
> '/usr/ports/distfiles/libreoffice/Python-3.5.0.tgz', needed by 
> '/usr/obj/ports/libreoffice-5.2.4.2/libreoffice-5.2.4.2/workdir/UnpackedTarget/Python-3.5.0.tgz'.
>   Stop.

So that one is weird, i had a look and Makefile says:

# XXX can't use ${MODPY_DEFAULT_VERSION_3}, it specifically wants 3.3
# and will try (and fail) to build its own copy if not available.
MODPY_VERSION=  ${MODPY_DEFAULT_VERSION_3}

but its not commented out ?

patch-configure tries to use/detect python3.4 but hardcodes it, maybe a
solution would be to add python${MODPY_DEFAULT_VERSION_3} to the patch
and run SUBST_CMD on configure ? That's what im going to try locally.
Robert, any idea ?

Landry

Reply via email to