[opensuse-packaging] scons is going to use python3 as interpreter
Dear Tumbleweed hackers and packagers, As you all know, python3 is 'the new thing' and pytho2 is being moved to the background. Various packages have already been switched to be using python3 - probably 'the easy ones' are all done. so we enter the jounery of more complex things. One such package is scons, which has been switched to be using python3 in request https://build.opensuse.org/request/show/542087 Now, scons is a bit 'special', as it is written in python and is basiclaly compatible to python2 AND python3, BUT it parses SConstruct/SConscript of different packages. And there the results are a bit mixed when it comes to python3 compliance Martin setup a test branch where he monitors all packages in openSUSE:Factory using scons as their build script, see https://build.opensuse.org/project/monitor/home:pluskalm:python3 Out of those only two packages are tracked as part of the rings (ffado and gpsd), which means from a Factory process perspective, once those two are fixed, the scons-switch to python3 could happen. As this will impact some other packages, I'd like to ask for your help to fix the other failing packages there. Simply branch them using "osc branch openSUSE:Factory $pkg" and submit them back to their original devel project. Martin's project only serves as monitor and should not receive submissions. In many cases, the issue in SConstruct are the usual python3 offenders: print "FOO" needs to be written as print ("FOO") now; so patches can turn out to be easy (in some cases they are more complex though) This mail serves as an info so you won't be surprised when scons is being merged and some packagaes stop building (of course some people will still be surprised, but such is life) Cheers Dominique
On 23/11/2017 15:13, Dominique Leuenberger / DimStar wrote:
Out of those only two packages are tracked as part of the rings (ffado and gpsd), which means from a Factory process perspective, once those two are fixed, the scons-switch to python3 could happen.
I've sorted out the rest of the print() errors, maybe I should submit it so long. I'm now stuck on a strange error: Checking for a working C-compiler TypeError : sequence item 0: expected str instance, bytes found Giving no indication of where the error is but a context.TryRun( to test the c compiler is the point that it's at. Unfortunately my scons and python2 to 3 skills are a bit rusty so any help will be appreciated. https://build.opensuse.org/package/show/home:plater/ffado I've failed to get a more verbose error so far. Thanks Dave P -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
Dave Plater
On 23/11/2017 15:13, Dominique Leuenberger / DimStar wrote:
Out of those only two packages are tracked as part of the rings (ffado and gpsd), which means from a Factory process perspective, once those two are fixed, the scons-switch to python3 could happen.
I've sorted out the rest of the print() errors, maybe I should submit it so long. I'm now stuck on a strange error: Checking for a working C-compiler TypeError : sequence item 0: expected str instance, bytes found Giving no indication of where the error is but a context.TryRun( to test the c compiler is the point that it's at. Unfortunately my scons and python2 to 3 skills are a bit rusty so any help will be appreciated. https://build.opensuse.org/package/show/home:plater/ffado I've failed to get a more verbose error so far.
This error is typically due to a mixup of raw bytes and unicode strings. In python 2, there were two types of strings: ascii strings and unicode strings. In practice, you'd read from a command line tool and get bytes which would be interpreted as an ascii string. In python 3, there are only unicode strings. All of the file reading or command line tool reading code now return bytes, that have to be decoded into unicode using a particular interpretation: ascii, utf-8, or something else. If the error had been reported with a file and line number, it'd be easier to figure out what was going wrong...
Thanks Dave P -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
-- // Kristoffer Grönlund // kgronlund@suse.com -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On Thu, 2017-11-23 at 15:28 +0200, Dave Plater wrote:
str instance, bytes found Giving no indication of where the error is but a context.TryRun( to test the c compiler is the point that it's at. Unfortunately my scons and python2 to 3 skills are a bit rusty so any help will be appreciated. Actually fedora solved this correctly - working version is at home:pluskalm:python3/ffado [1] (fixed by tacit).
Cheers M 1. https://build.opensuse.org/package/rdiff/home:pluskalm:python3/ffado?linkrev=base&rev=2 -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 24/11/2017 11:18, Martin Pluskal wrote:
On Thu, 2017-11-23 at 15:28 +0200, Dave Plater wrote:
str instance, bytes found Giving no indication of where the error is but a context.TryRun( to test the c compiler is the point that it's at. Unfortunately my scons and python2 to 3 skills are a bit rusty so any help will be appreciated. Actually fedora solved this correctly - working version is at home:pluskalm:python3/ffado [1] (fixed by tacit).
Cheers
M
1. https://build.opensuse.org/package/rdiff/home:pluskalm:python3/ffado?linkrev=base&rev=2
Actually Fedora solved this with: + return True # FIXME: The following TryRun fails on Python3 I don't call that a fix I call it a work around. Dave P -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On Fri, 2017-11-24 at 11:58 +0200, Dave Plater wrote:
On 24/11/2017 11:18, Martin Pluskal wrote:
On Thu, 2017-11-23 at 15:28 +0200, Dave Plater wrote:
str instance, bytes found Giving no indication of where the error is but a context.TryRun( to test the c compiler is the point that it's at. Unfortunately my scons and python2 to 3 skills are a bit rusty so any help will be appreciated.
Actually fedora solved this correctly - working version is at home:pluskalm:python3/ffado [1] (fixed by tacit).
Cheers
M
1. https://build.opensuse.org/package/rdiff/home:pluskalm:python3/f fado?linkrev=base&rev=2
Actually Fedora solved this with: + return True # FIXME: The following TryRun fails on Python3
I don't call that a fix I call it a work around.
Well call it as you wish but apparently Fedora had already succesfully switched scons to python3 and has building packages :) Cheers M -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
participants (4)
-
Dave Plater -
Dominique Leuenberger / DimStar -
Kristoffer Grönlund -
Martin Pluskal