[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gpsd-dev] report on my build nits
From: |
Greg Troxel |
Subject: |
Re: [gpsd-dev] report on my build nits |
Date: |
Mon, 23 Sep 2019 19:53:13 -0400 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/26.1 (berkeley-unix) |
Fred Wright <address@hidden> writes:
>>> Easy. Use the "target_python=python3.7" scons config option.
>>
>> I did that, and the test suite had problems, I think because
>> target_python didn't get used for the tests that are written in python.
>
> It *does* get used for the tests, and I see no evidence to the
> contrary. Note, however, if you 'build' and 'check' with different
> target_python settings, it will probably fail to notice that the
> extensions were built for the wrong Python version, and thus won't
> work properly. You'd need to go out of your way to do that, though,
> since the options are sticky.
I don't follow this. There are tests with "#!/usr/bin/env python"
(which isn't the value of target_python), and running them qfails because
there is no bare python. If there were a bare python, it would be
executing some python version which might or might not be the same as
target_python, which to me feels like undefined behavior.
So separately form arguing about following PEPfoo etc., I think it's
good to test the machinery at times on a system with no bare python,
because I think the gpsd build invoking bare python (when given
target_python=python3.7) is a bug.
I tried to build with target_python=python3.7 and then assumed check
would use the cached value, since scons seems to be quite aggressive
about using cached values on future runs. Perhaps that was a mistake on
my part.
Fair enough that changing target_python between build and check should
fail, as absolutely it should be loading the python code from the build
and not loading anything from the installed system.
> The only real bug I'm aware of is that the default target_python
> doesn't work if your system has no 'python' command. I'll look at
> fixing that. Note that an explicitly blank target_python uses whatever
> Python is being used by SCons.
Thanks - that would be great!
- Re: [gpsd-dev] report on my build nits, (continued)
- Re: [gpsd-dev] report on my build nits, Bernd Zeimetz, 2019/09/23
- Re: [gpsd-dev] report on my build nits, James Browning, 2019/09/23
- Re: [gpsd-dev] report on my build nits, Bernd Zeimetz, 2019/09/23
- Re: [gpsd-dev] report on my build nits, James Browning, 2019/09/23
- Re: [gpsd-dev] report on my build nits, Gary E. Miller, 2019/09/23
- Re: [gpsd-dev] report on my build nits, Greg Troxel, 2019/09/23
- Re: [gpsd-dev] report on my build nits, Gary E. Miller, 2019/09/23
- Re: [gpsd-dev] report on my build nits, Fred Wright, 2019/09/23
- Re: [gpsd-dev] report on my build nits,
Greg Troxel <=
- Re: [gpsd-dev] report on my build nits, Fred Wright, 2019/09/23
- Re: [gpsd-dev] ✘timestamp_t is dead. Long live timespec_t, Gary E. Miller, 2019/09/20
- Re: [gpsd-dev] ✘timestamp_t is dead. Long live timespec_t, Greg Troxel, 2019/09/20
- Re: [gpsd-dev] ✘timestamp_t is dead. Long live timespec_t, Gary E. Miller, 2019/09/20
- Re: [gpsd-dev] ✘timestamp_t is dead. Long live timespec_t, Greg Troxel, 2019/09/20
- Re: [gpsd-dev] ✘timestamp_t is dead. Long live timespec_t, Gary E. Miller, 2019/09/20
- Re: [gpsd-dev] ✘timestamp_t is dead. Long live timespec_t, Greg Troxel, 2019/09/21