gpsd-dev
[Top][All Lists]
Advanced

[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: Fri, 20 Sep 2019 21:28:07 -0400
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/26.1 (berkeley-unix)

"Gary E. Miller" <address@hidden> writes:

>> but
>> gpsd should be able to cope with PYTHON naming the one to use when
>> there is no python.  It turns out it basically can, with another
>> variable.
>
> Yup.  Well, several of them...
>
>> >> The docs didn't explain the scons target_python option,
>> >> needed to find a python interpreter not named python.  
>> >
>> > You can't just change the name of the python program, you also,
>> > usualy, need to change the PYTHONPATH.  
>> 
>> Each python program has a compiled-in default PYTHONPATH that works;
>> python3.7 e.g. looks in
>> /usr/pkg/lib/python3.7/site-packages/./usr/pkg/lib/python3.7/site-packages/
>
> Wow, that is weird.  Never seen that.  Why would someone put usr/pcg
> under another /usr/pkg?

sorry, paste error, meant just

  /usr/pkg/lib/python3.7/site-packages

> Either way, it is sorta circular, you need the PYTHONPATH to find the
> PYTHONPATH in those files...

I don't follow.  import just looks there.

>> > Eric keeps insisting no the current behavior.  It does simplify a
>> > lot of things and has quickly identified bad installs.  
>> 
>> It does, but it also breaks the notion of gpsd as a network protocol
>> where you can run a client on one machine and the gpsd server on
>> another.
>
> Only when they are different versions.  You'll need to take that up with
> Eric.

EDONTTILTATWINDMILLS

>>  Arguably there should be a protocol version instead, with
>> significant effort not to have a breaking change.
>
> Already done.  This is what it looks like:
>
> # gpspipe -w pi8
> {"class":"VERSION","release":"3.19.1~dev","rev":"release-3.19-527-g670e7c538","proto_major":3,"proto_minor":14}
>
> Notice the VERSION and rev.

OK, but I really meant further that the client/server check should be on
protocol version.

Imagine if Thunderbird 60 couldn't talk to dovecot 2.3 because the
prgoram versions are different.

Anyway - I've given my opinion and am now done, until I forget and run
into this again, when I'll mention it again :-)



reply via email to

[Prev in Thread] Current Thread [Next in Thread]