This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: real_selected_kind behavior

From: Jerry DeLisle <jvdelisle at frontier dot com>
To: Thomas Koenig <tkoenig at netcologne dot de>
Cc: Tobias Burnus <tobias dot burnus at physik dot fu-berlin dot de>, Kyle Horne <horne dot kyle at gmail dot com>, fortran at gcc dot gnu dot org
Date: Wed, 27 Apr 2011 18:44:16 -0700
Subject: Re: real_selected_kind behavior
References: <20110427153223.GA3908@physik.fu-berlin.de> <4DB8766B.4060003@netcologne.de>

On 04/27/2011 01:02 PM, Thomas Koenig wrote:

Am 27.04.2011 17:32, schrieb Tobias Burnus:

One could partially solve the problem by moving REAL_SELECTED_KIND into
the front end (which is a lot effort), but it wouldn't solve the problem
of the configration dependency as a real-16 capable compiler could still
link against a real-16 incapable library.


If we really want to be sure, then we need to check both the front end and the
library - if either of them doesn't support a certain precision, then return -1
as kind.

Thomas

Would it be possible for us to encode into the options structure information about the frontend and have the library check it at run time and either give an error or warning? For example, we could return the front-end version number to the library and have the library simply check that the library is not ahead or behind. Also, it seems we could also pass a vector of supported kinds from the front-end so the library could report a different result.

Just some thoughts.

Jerry

References:
- Re: real_selected_kind behavior
  - From: Tobias Burnus
- Re: real_selected_kind behavior
  - From: Thomas Koenig

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]