Re: [PATCH] ltmain.in: don't suppress output for PIC compilations

[Ileana Dumitrescu]

On 15/08/2024 20:37, Sam James wrote:
> Nick Bowler <nbowler@draconx.ca> writes:
>
> > On 2024-08-15 12:13, Sam James wrote:
> > > Ileana Dumitrescu <ileanadumitrescu95@gmail.com> writes:
> > > > However, you could also just specify '-no-suppress' when compiling
> > > > rather than changing the default behaviour.
> > >
> > > (Is there a way to do this globally? If I put it in CFLAGS globally,
> > > won't this break if libtool isn't used for a package, given GCC or Clang
> > > will see it and bail out?)
> >
> > Automake provides LIBTOOLFLAGS and AM_LIBTOOLFLAGS for this kind of
> > purpose but unfortunately it is not in the right position to pass
> > mode-specific options like -no-suppress.
> >
> > I think it would be a big improvement to allow most mode-specific options
> > to appear earlier on the libtool command line.  Then you could just do
> >
> >    make LIBTOOLFLAGS=-no-suppress
> >
> > and it'd work.
>
> Ah, thanks. This is a good idea.


There is a relevant bug filed in Automake discussing this:
https://debbugs.gnu.org/cgi/bugreport.cgi?bug=54020

> > > I really feel like the current behaviour is unexpected and surprising,
> > > and perhaps it was based on the (wrong) idea that the -fPIC vs non-PIC
> > > build will produce identical warnings and if one fails, the other will
> > > fail, but that's not true at least with modern compilers.
> >
> > Can we fix the compilers instead?  Why does PIC compilation suppress
> > warnings?  That behaviour seems unexpected and surprising.
>
> Because PIC can substantially affect generated output. Not only that, as
> in the GCC bug I linked to, this behaviour hid a compiler crash.
>
> The same thing could be true if an error is only emitted without PIC.
>
> > > That's the part I want to discuss -- does anyone actually think the
> > > suppression is still a good idea?
> >
> > Duplicate output sounds very annoying so yes, I think it is a good idea
> > to suppress duplicate output.
>
> More annoying to have an error suppressed that is not obvious to debug,
> especially if one isn't aware of this behaviour.
>
> > If you change the default, maybe just print the differences on the
> > second run.  This might be as simple as just running diff on the
> > output of both runs.
>
> Yeah, that might be a good compromise.

Printing the differences on the second run seems like a better default,
while keeping the option for -no-suppress for the complete log. Maybe
adding an informational message about -no-suppress with the diff
would help resolve some confusion?

However, I also wonder if the suppression should be decided by a
verbosity option/variable rather than a compile mode option.

> > Cheers,
> >    Nick

--
Ileana Dumitrescu

GPG Public Key: FA26 CA78 4BE1 8892 7F22 B99F 6570 EA01 146F 7354

OpenPGP_0x6570EA01146F7354.asc
Description: OpenPGP public key

OpenPGP_signature.asc
Description: OpenPGP digital signature