[Scons-dev] Help needed on command string

[Bill Deegan]

Russel,

I'm not sure what the difference is between configurations and variants?
To me they are the same and SCons supports them well.
The only difference may be that other build systems have per-concieved
ideas of different configurations (flags for optimize/release/debug/etc)
(More easily for some directory structures than others)

Typically you create a bunch of environments for each variant, and then
pass that to the same SConscript with different variant_dir's.

In some cases you may use info passed from the parent (Via SConscript) to
modify the behavior of the children (SConscripts).

I've built debug/release/test/purify X 32/64bit x {linux,win32, solaris,
hpux, aix} with such a build system. Worked like a charm.  (And wasn't
overly complicated)

-Bill

On Wed, Apr 12, 2017 at 3:04 AM, Russel Winder <russel at winder.org.uk> wrote:

> On Tue, 2017-04-11 at 12:47 -0700, Bill Deegan wrote:
> > D tools are doing nothing wrong.
> > The same thing would happen with any compiler.
> >
>
> Phew. :-)
>
> Good to know, thanks. I'll ditch what I was assembling as an error test
> case.
>
> However, even though the VariantBuild stuff is working as those who
> know how it works, is it the way it should be?
>
> I believe that D is bringing a new use case to the table, one that
> disturbs the SCons model.
>
> With the C++ codes that used to have a SCons build (sorry but I had to
> switch them to CMake and Meson :-) I never had a problem with the
> variant_dir in the SConscript call. It worked as I expected. It turns
> out though that my mental model may not have been matching the reality.
>
> I am not sure I full understand Bill's fix for my ApproxGC SCons build,
> but it looks ungainly. The new problem is that the same file has to be
> compiled in two different ways, one for the application and once for
> the tests. C, C++, Fortran do not have this situation. CMake and Meson
> do not have a problem with this because they decouple each target, the
> build rules for each target are totally separate. Also CMake and Meson
> distinguish targets and variants, i.e. the release, debu, etc. variant
> is totally separate from how to build the targets. SCons tends to merge
> these by using the Variants for everything.
>
> I am assuming that the history of SCons working with C, C++, Fortran
> means that the variant systems has never had to deal with two builds of
> the same file as "configurations" rather than "variants". Is there
> something that needs addressing here?
>
> --
> Russel.
> ============================================================
> =================
> Dr Russel Winder      t: +44 20 7585 2200   voip:
> sip:russel.winder at ekiga.net
> 41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel at winder.org.uk
> London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder
>
> _______________________________________________
> Scons-dev mailing list
> Scons-dev at scons.org
> https://pairlist2.pair.net/mailman/listinfo/scons-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist2.pair.net/pipermail/scons-dev/attachments/20170412/1fdd49d6/attachment.html>