Re: Re: How is it possible ???
Alec <[email protected]>
| Newsgroups | gmane.linux.gentoo.performance |
|---|---|
| Message-ID | <[email protected]> |
Adam Petaccia wrote: >On Mon, 2005-01-10 at 21:55 -0500, Ted Rodgers wrote: > > >>>On Thursday 06 January 2005 04:46, William Kenworthy wrote: >>> >>> >>>>Have a read of the ebuild - compiling gentoo with your own choice of >>>>optimisations is is almost a myth! There's been some interesting >>>>threads about the predominance of flag filtering and how it often seems >>>>to be used with little cause. >>>> >>>> >>>I have a hacked gcc-config profile that calls a perl script for all the >>>gcc >>>executables that forces my cflags to be what I want and aside from ten >>>packages on my entire system I'm able to run without any problems. That >>>includes X (though not glibc). I think strip-compiler flags might be a >>>good >>>idea for some systems but clearly on mine it's not necessary. >>> >>>-- >>>You can get so much more with a kind word and a two-by-four than you >>>can get with just a kind word. >>>... >>>Patrick Audley [email protected] >>>Blackcat Systems http://blackcat.ca >>> Bringing Elegance to Complexity >>> >>> >>If you read the dev docs, it is possible to stop portage from stripping >>flags. >> >> >FEATURES=nostrip USE=debug emerge foo > > > USE=debug is problematic...Emerging certain packages (nano for example) with USE=debug breaks the program completely, mostly due to the fact that the DEBUG flag does non-standard things. Some packages append -g to the CFLAGS with use debug, some just add in --enable-debug to configure, and yet others do other weird stuff. I know there was a discussion about the DEBUG flag and how crappy it is on the -dev list, it may be a hack, but it's certainly not a 'supported' or good way to stop the stripping of flags. >>Ted >> >> >> >> >>>-- >>>[email protected] mailing list >>> >>> >>> >>> >>-- >>[email protected] mailing list >> >> >> -- Alec Warner Spartasoft Secretary ( spartasoft.msu.edu ) Junior Computer Science Michigan State University [email protected] -- [email protected] mailing list