Andrew Pinski - Re: [fortran] Disable use of -malign-double (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: Andrew Pinski
• To: tobias dot schlueter at physik dot uni-muenchen dot de (Tobias Schlüter)
• Cc: pinskia at physics dot uc dot edu (Andrew Pinski), sgk at troutmask dot apl dot washington dot edu (Steve Kargl), jvdelisle at verizon dot net (Jerry DeLisle), fortran at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org
• Date: Thu, 1 Feb 2007 04:08:15 -0500 (EST)
• Subject: Re: [fortran] Disable use of -malign-double

Andrew Pinski wrote:

So that is their problem, when stuff does not work and they use an option which changes the ABI, then they get what they deserve. Really we should not have these options but for somereason in the past, people thought it was a good idea to expose stuff like this.

I don't think we can assume that our users know what an "ABI" is. They read the documentation, read "faster code", and since they compile all their code themselves(*) they don't worry, about the part where it says that it may not work to combine code that is in part compiled with -malign-double and in part not.

Really they should not be using flags that they don't understand. There is a big warning already in the documentation about the option, that should be enough.

If it's at all possible the right fix would be to make sure that library calls don't depend on whether the user explicitly gave -malign-double or not. For the time being, I agree with Steve that -malign-error should give a hard error iff it's not the system's default.

How do you prove it is not the system's default? It is a hard thing. Again I can go and make a GNU/Linux distro, where the GCC's specs file passes in -malign-double and have it work. I am not editting any other file really. And now with the proposed patch, it will not work.

Then -malign-double is the default for your system, and using it would indeed not constitute an error.

How do you prove it is default? As shown above it is hard to prove it. Also this is a target specific flag and should not make a difference to the front-ends. Any option starting with -m should not be in a lang.opt.

I hope the gfortran people realize it is not about erroring out but really about having a clearer line between the front-end and the back-end(target).

Thanks, Andrew Pinski

• Follow-Ups:
  • Re: [fortran] Disable use of -malign-double
    * From: Tobias Schlüter
• References:
  • Re: [fortran] Disable use of -malign-double
    * From: Tobias Schlüter