New Build System Status

Wed Nov 27 11:49:55 UTC 2019

Hi Sebastian,

Thanks for that great effort. I'd aim to use this build system for my
RISC-V development.

I followed the user manual trying to build RISC-V targets and RTEMS
(aaf7f8b84) and here are a few comments:

* .waf bsp_defaults doesn't give an error when mistyping the BSP name,
but just outputs an empty .ini file.
* ./waf complains about not finding gcc if I don't give it the
--prefix, even though it's in my PATH
* dltests fail to build on RISC-V when BUILD_TESTS = True
* Can't find  any "*.exe" after a default build:
* Can't notice a difference when I pass rtems-compiler=clang, it still
builds with gcc.

./waf bsp_defaults --rtems-bsps=riscv/rv64imac_medany  > bsps.ini
./waf configure
./waf

On Wed, 27 Nov 2019 at 10:01, Sebastian Huber
<sebastian.huber at embedded-brains.de> wrote:
>
> On 27/11/2019 00:59, Chris Johns wrote:
> > On 27/11/19 2:07 am, Sebastian Huber wrote:
> >> I updated the build system documentation.
> >
> > Thank you.
>
> Thanks for having a look at it and the very helpful comments.
>
> >
> >> It is now sufficiently good to get integrated from my point of view.
> >
> > Excellent and well done. However it cannot be merged until the question I raised
> > here is resolved ...
> >
> > https://lists.rtems.org/pipermail/devel/2019-November/056225.html
> >
> > Joel raised it and it is valid so I feel we should to wait until he responds. Joel?
> >
> >> https://ftp.rtems.org/pub/rtems/people/sebh/eng.pdf
> >
> > This is a really great start piece of work. To have this from the start is so good.
> >
> > 8.1 Should the install location be specified rather than referencing something
> > that is not specified?
> >
> > "Configurable things which define what is built and how the artefacts are
> > configured are intended to be placed in configuration files."
> >
> > to
> >
> > " Configurable things which define what is built and how the artefacts are
> > configured are intended to be placed in configuration files that can be
> > configuration controlled."
> >
> > "The wscript file should know nothing ..."
> >
> > to
> >
> > "The waf build script file called wscript should know nothing ..."
>
> Ok, I reworded this.
>
> > 8.2
> >
> > What is the link to "RTEMS User Manual"? I think inter-document links are
> > problematic. Referencing the docs.rtems.org can result in version mismatches.
>
> I think links to other documents are quite helpful. We should find a way
> to make them happen. For example a script could adjust the URLs in a
> release branch from the "master" to the right version. Maybe add the
> manuals to
>
> common/rtemsdomain.py
>
> and reference chapters/sections by name?
>
> >
> > Should there be a simple overview sequence picture ...
> >
> >    -----------------------------
> >            get defaults
> >    -----------------------------
> >                 |
> >                 +-- edit --+
> >                 |          |
> >    -----------------------------
> >             configure
> >    -----------------------------
> >                 |
> >    -----------------------------
> >               build
> >    -----------------------------
> >                 |
> >    -----------------------------
> >              install
> >    -----------------------------
> >
> > ... as I think it would help explain the next sections?
>
> I think this is already covered by the User Manual content.
>
> >
> > 8.3 Please add an example of the command. I know what you mean however I suspect
> > I am one of a few who would.
>
> Ok, done.
>
> >
> > 8.5
> >
> > Order: In the trade off case how does someone get a current picture of this
> > structure so they can determine a suitable value?
>
> Ok, I restructured this section.
>
> >
> > cflags: Can any valid cflags be added here? Are there constraints on cflags and
> > any of the other types of flags listed?
>
> Currently, no variable substitution is performed on the flags. This is a
> fine tuning option. We should measure how much performance the
> substitution would cost.
>
> > The example uses [], is this a valid
> > YAML syntax?
>
> Yes, of course. It denotes an empty list.
>
> >
> > 8.5.2 The cflags page reference is wrong, I see cflags on page 118 and not 116.
>
> Hm, this looks like a Sphinx/Tex bug. The figure moves into the space
> between the label and the content.
>
> .. _BuildAttrCflags:
>
> cflags
>      The `cflags` value shall be a list of options for the C compiler.
>
> I don't know how to fix this.
>
> >
> > format: Does a valid python string mean suitable indenting? Does it need to meet
> > any required indent level or is in self contained?
>
> I added an example.
>
> >
> > name: Any constraints such as valid characters that can be used?
>
> I added a regular expression.
>
> >
> >> https://ftp.rtems.org/pub/rtems/people/sebh/user.pdf
> >
> > Nice work.
> >
> > The last step does another `cd` to the RTEMS source, could this be confusing?
>
> I think the boxes should be self-contained, e.g. if a command needs to
> be called in a certain directory, then this should be ensured.
>
> >
> > 7.3 More external links. Same issue as the eng manual, I cannot see how these
> > can be made to work for any build of documentation at any location for the
> > various types of output we generate.
>
> See above.
>
> --
> Sebastian Huber, embedded brains GmbH
>
> Address : Dornierstr. 4, D-82178 Puchheim, Germany
> Phone   : +49 89 189 47 41-16
> Fax     : +49 89 189 47 41-09
> E-Mail  : sebastian.huber at embedded-brains.de
> PGP     : Public key available on request.
>
> Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
> _______________________________________________
> devel mailing list
> devel at rtems.org
> http://lists.rtems.org/mailman/listinfo/devel