* [gentoo-user] Calls compiler directly, and strange parameters passed. Is this a bug?
@ 2011-10-29 0:03 Michael Mol
2011-10-31 9:40 ` James Broadhead
0 siblings, 1 reply; 2+ messages in thread
From: Michael Mol @ 2011-10-29 0:03 UTC (permalink / raw
To: gentoo-user
Just update calibre, and it looks like it builds a bunch of Python
extensions. In the build output, I see many things like these:
g++ -O3 -Wall -DNDEBUG -fno-strict-aliasing -pipe -fPIC
-march=amdfam10 -mcx16 -msahf -mpopcnt --param l1-cache-size=64
--param l1-cache-line-size=64 --param l2-cache-size=512
-mtune=amdfam10 -O2 -pipe -ggdb -pthread -I/usr/include/python2.7
-DPNG_SKIP_SETJMP_CHECK -DPOPPLER_OLD_LINK_TYPE -I/usr/include
-I/usr/include/ImageMagick -I/usr/include/poppler -c
/var/tmp/portage/app-text/calibre-0.8.23/work/calibre/src/calibre/ebooks/pdf/images.cpp
-o /var/tmp/portage/app-text/calibre-0.8.23/work/calibre/build/objects/pdfreflow/images.o
gcc -O3 -Wall -DNDEBUG -fno-strict-aliasing -pipe -fPIC
-march=amdfam10 -mcx16 -msahf -mpopcnt --param l1-cache-size=64
--param l1-cache-line-size=64 --param l2-cache-size=512
-mtune=amdfam10 -O2 -pipe -ggdb -pthread -I/usr/include/python2.7
-I/usr/include/ImageMagick -c
/var/tmp/portage/app-text/calibre-0.8.23/work/calibre/src/calibre/utils/magick/magick.c
-o /var/tmp/portage/app-text/calibre-0.8.23/work/calibre/build/objects/magick/magick.o
For reference, here are my CFLAGS:
CFLAGS="-march=amdfam10 -mcx16 -msahf -mpopcnt --param
l1-cache-size=64 --param l1-cache-line-size=64 --param
l2-cache-size=512 -mtune=amdfam10 -O2 -pipe -ggdb"
CXXFLAGS="${CFLAGS}"
It looks like the package adds "-O3 -Wall -DNDEBUG
-fno-strict-aliasing -pipe -fPIC" before appending CFLAGS or CXXFLAGS
to the line, and it's calling 'gcc' or 'g++' directly, rather than
x86_64-pc-linux-gnu-gcc, x86_64-pc-linux-gnu-g++ or something more
generic. It was my understanding that the 'right' way to create a
build system is to use the compiler for the target platform. I *think*
that means the commands I noted, but I'm uncertain
(And there's that it uses -O3, which is a bit spooky, and makes me
worry more about the effects of compiler upgrades)
Is this incorrect? Does it qualify as a bug?
--
:wq
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [gentoo-user] Calls compiler directly, and strange parameters passed. Is this a bug?
2011-10-29 0:03 [gentoo-user] Calls compiler directly, and strange parameters passed. Is this a bug? Michael Mol
@ 2011-10-31 9:40 ` James Broadhead
0 siblings, 0 replies; 2+ messages in thread
From: James Broadhead @ 2011-10-31 9:40 UTC (permalink / raw
To: gentoo-user
On 29 October 2011 01:03, Michael Mol <mikemol@gmail.com> wrote:
> Is this incorrect? Does it qualify as a bug?
Yes. The ebuild should respect your CFLAGS unless there is a USE flag
to enable/disable them. The counter to this is when a package won't
build without specific flags, then disregarding the user is okay.
Since this is over-optimising (rather than under-), this probably
isn't the case.
File a minor bug, and if you're super-interested, please take a look
at ebuild writing. I get the feeling that maintaining calibre is a bit
messy, I'm sure that they'd appreciate the help.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2011-10-31 9:42 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-10-29 0:03 [gentoo-user] Calls compiler directly, and strange parameters passed. Is this a bug? Michael Mol
2011-10-31 9:40 ` James Broadhead
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox