CFLAGS, C++, -frepo reduces size by 25%

colonel_dolphin · n00b Joined: 12 Jan 2004 Posts: 39

According to a test by the ACE/TAO group, the gcc option "-frepo" was reported to reduce footprint of statically built ACE/TAO by 25%.

Using "-frepo" in CXXFLAGS reduces the footprint, but increases the compile time.

Does anyone have interesting benchmarks (file size, RSS, CPU) comparing with and without "-frepo" for a real-world application, like htdig?

The latest version of gcc I found seems to imply that -frepo is still usefull. However, I also found mention that newer version of the GNU toolchain (i.e. ld) is capable of automatically removing the duplication resulting if "-frepo" is not used. Hmmmm.
http://gcc.gnu.org/onlinedocs/gcc-3.4.0/gcc/Template-Instantiation.html#Template%20Instantiation

Halcy0n · Posted: Mon Jun 07, 2004 10:56 pm Post subject:

Put -frepo into your CXXFLAGS, those should be used when just compiling C++.
CXXFLAGS = "{$CFLAGS} -frepo"
_________________
Mark Loeser
http://www.halcy0n.com

asph · Posted: Fri Sep 24, 2004 11:47 am Post subject:

-frepo breaks ncurses-5.4-r5
_________________
gentoo sex is updatedb; locate; talk; date; cd; strip; look; touch; finger; unzip; uptime; gawk; head; emerge --oneshot condom; mount; fsck; gasp; more; yes; yes; yes; more; umount; emerge -C condom; make clean; sleep

Torangan · Apprentice Joined: 21 Mar 2003 Posts: 178

You must have done something strange. ncurses is a C program and not influenced by CXXFLAGS. Therefore it did compile again without problems here after I added -frepo to my CXXFLAGS.

Xaignar · Posted: Fri Sep 24, 2004 6:21 pm Post subject:

Well, I can tell you that a CVS build of aMule fails if you use that CXXFLAG, cause i just tried it. Caused a cartload of linking errors related to embedded Crypto++ code.

Genkaku · n00b Joined: 26 Aug 2004 Posts: 72 Location: Poland

We can read in point 3 of options that are present.:

Do nothing. Pretend G++ does implement automatic instantiation management. Code written for the Borland model will work fine, but each translation unit will contain instances of each of the templates it uses. In a large program, this can lead to an unacceptable amount of code duplication.

That should explain a little :-)

Gavinv · n00b Joined: 05 Aug 2004 Posts: 10

Hmmm, so perhaps what we really need is a matrix of valid compile options for specific platforms / ebuilds?

https://bugs.gentoo.org/show_bug.cgi?id=57223

Bowyakka · Posted: Wed Feb 23, 2005 4:48 pm Post subject:

This seems ok but breaks fam and doxygen
_________________
glows in the dark

Dr.Dran · Posted: Sat Dec 31, 2005 12:26 pm Post subject:

Excuse me I resume this post because I would like to know if the -frepo option for the current stable gcc-3.4.4-r1 is buggy or safe to turn on in a production system.

Ciauz and have a good hacking

_________________
:: [Dr.Dran] Details ::
- Linux User # 286282
- IT FreeLance Consultant
- President of ImoLUG [Imola & Faenza Linux User Group]

ZeuZ_NG · Posted: Wed Jan 28, 2009 7:47 am Post subject:

BUMP!
As of 2008, how has this flag evolved?
_________________
| Intel Core i7 920. | Intel DX58SO Extreme. |
| 8gb DDR3 1666mHz | 1TB 7200 RPM WD HDD.|
Unix´s guru view of sex:
unzip ; strip ; touch ; grep ; finger ; \
mount ; fsck ; more ; yes ; umount ; sleep.