On Thu, Jul 03, 2003, julien Touche wrote:

> >Err... as I said: "...as source (principle of pristine source files) and
> >intentionally do not provide the enabling of the patch's functionality
> >by default (because this breaks POLA IMHO too much)." If you want other
> >packages built with gcc and this option enabled, you can use --define
> >"use_cc /prefix/bin/gcc -f...." on the RPM command line (if that is what
> >you want to achieve).
>
> maybe i misunderstood here, but POLA (have a link ? i don't find it) ask
> to have no extra option enabled by default (right?).

No, sorry for using this acronym: POLA = Principle Of Least
Astonishment. See for instance this URL (one of many Google gave out to
me now) http://paulbeard.no-ip.org/movabletype/archives/000048.html for
some more details. Sorry that I've described it not clear enough.

My point just is that building GCC and have such a compiler option
enabled by default might lead to more astonishment by the user than
expected, because usually you use the GCC -fxxx options to explicitly
enable something (although there exists also -fno-xxx for some things
which are enabled by default, but those are all standard and known
features while the SSP stuff is not part of this group IMHO). Hence I
was not a fan of having an OpenPKG package option for supporting this.
I think the SSP stuff (like the bounds stuff) should be something which
is optionally built into GCC and has to be explicitly enabled under
run-time by the end-user.
                                       Ralf S. Engelschall
                                       [EMAIL PROTECTED]
                                       www.engelschall.com

______________________________________________________________________
The OpenPKG Project                                    www.openpkg.org
User Communication List                      [EMAIL PROTECTED]

Reply via email to