RE: [OE-core] [PATCH] gcc: Allow -Wno-error=poison-system-directories to take effect

[Peter Kjellerstedt <peter.kjellerstedt@axis.com>]

> -----Original Message-----
> From: Khem Raj <raj.khem@gmail.com>
> Sent: den 13 oktober 2022 17:41
> To: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
> Cc: Ross Burton <Ross.Burton@arm.com>; Patches and discussions about the
> oe-core layer <openembedded-core@lists.openembedded.org>
> Subject: Re: [OE-core] [PATCH] gcc: Allow -Wno-error=poison-system-
> directories to take effect
> 
> On Thu, Oct 13, 2022 at 7:38 AM Peter Kjellerstedt <peter.kjellerstedt@axis.com> wrote:
> >
> > > -----Original Message-----
> > > From: Ross Burton <Ross.Burton@arm.com>
> > > Sent: den 13 oktober 2022 16:05
> > > To: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
> > > Cc: Patches and discussions about the oe-core layer <openembedded-
> > > core@lists.openembedded.org>
> > > Subject: Re: [OE-core] [PATCH] gcc: Allow -Wno-error=poison-system-
> > > directories to take effect
> > >
> > > Whilst this is a good fix, I’m horrified to ask why you need it…
> >
> > We have a recipe for skia. Somewhere in that mess of a build
> > system it uses (gn), something adds -I/usr/include/freetype2,
> > which results in an error with -Werror=poison-system-directories
> > enabled (as it should). Now, to work around that problem, the
> > recipe currently adds -I=/usr/include/freetype2, which means we
> > know that the correct path is used. However, since the original
> > -I/usr/include/freetype2 is still in the list of options, the
> > error still triggers.
> >
> > Now, to fix the problem properly, skias's build files should be
> > patched, but that is not my job. What I am working on is to get
> > our recipes to build with Langdale. This means I can cut some
> > corners and, e.g., disable errors and then hand off to the
> > responsible team to fix the underlying problem at a time that
> > suits them. Thus in this case I wanted to add
> > -Wno-error=poison-system-directories for skia, which I then, to
> > my surprise, noticed had no effect...
> 
> from gcc recipe point of view, this fix is fine. however your use of
> this seems to be a workaround that perhaps is better fixed in skia

Oh, I'm well aware of that. As I said, my job at this stage is to get 
everything building with Langdale. And to achieve that, it is perfectly 
fine for me to, e.g., disable compiler errors. Once the platform 
builds, the respective code owners have the responsibility to actually 
correct their code to not generate the errors in the first place.

> because = syntax for sysroot works with gcc but may not work with
> clang e.g. so you are being lucky here.

We don't use clang so that is not a problem for us.

> > //Peter
> >
> > > Ross

//Peter