[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gpsd-dev] scons check fail gpsd3.14 ubuntu 14
From: |
Eric S. Raymond |
Subject: |
Re: [gpsd-dev] scons check fail gpsd3.14 ubuntu 14 |
Date: |
Tue, 17 Mar 2015 13:31:07 -0400 |
User-agent: |
Mutt/1.5.23 (2014-03-12) |
Hal Murray <address@hidden>:
>
> >> Shouldn't scons check use libgpsd.so.22.0.0 from the build directory?
> > You have to use chrpath=yes for that - this is explained in the build
> > instructions.
>
> If scons check really needs chrpath, it should bail with a sensible error
> message rather than testing with the wrong library which gives obscure error
> messages, or worse, actually works when the new library introduced a bug that
> scons check should have uncovered.
No, by that I mean that chrpath=yes is required to run scons check *using
libraries in the build directory*, not that it is required for scons check
to run at all.
> But I thought the plan for no chrpath was for scons check to use the local
> libraries and scons install to relink to use wherever they get installed.
Backwards. That's what happens with chrpath=yes. With chrpath=no the
binaries are built with an RPATH requiring the libraries to be in libdir,
in rootspace.
scons check works for me in this configuration. It doesn't work for you.
I still don't know why it doesn't.
> How hard would it be to add a switch so scons check could check the post
> install code?
Check it for what?
Greg Troxel's proposal - require either chrpath or static linking for scons
check - is looking better all the time. All this futzing around with edge
cases in the build is making me tired.
--
<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>