Re: ✘buildtmp

[Gary E. Miller]

Yo Fred!

On Mon, 2 Nov 2020 23:56:24 -0800 (PST)
Fred Wright <fw@fwright.net> wrote:

> > SConstruct tells scons to make a new directory, buildtmp/, and then
> > proceed to run SConscript inside buildtmp/.  This uses the
> > variant_dir features of scons.  Cleaning still needs to be cleaned
> > up, but is a simple as:
> >
> >        rm -r buildtmp/  
> 
> As long as one is working in a git repo, "git clean -dxf" has always
> been much faster and more thorough than anything involving scons.

Which only worked if you were pulling from git.  Most gpsd builders
build from the tarball.

The new way means .gitignore is simple and a bunch of gnarly env.Clean()
calls will go away.

I love it when a large change leads to many fewer LOC.

> > Everything runs and checks, for me.  There is some remaining
> > ugliness in the tests, which is a result of existing ugliness in
> > the tests.
> >
> > I also see the Debian CI pipeline failed.  I'll fix that shortly.
> >
> > Comments welcome, please test.  
> 
> Based on a quick check, I see that "build-all check" works but "www"
> fails miserably.

I should have been clear that only one build path had been tested.
Today all the CI jobs also work.

Yeah, all the stuff on the edges needs to be tested and adjusted. "www"
is an important one, but also the release testing, release building,
regression building and more.  But I think the hardest parts are done.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        gem@rellim.com  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

pgpbbmUK9nM7o.pgp
Description: OpenPGP digital signature