On Sat, 2003-01-25 at 12:14, Pigeon wrote: > On Sat, Jan 25, 2003 at 04:43:44PM +1100, Rob Weir wrote: > > On Thu, Jan 23, 2003 at 03:22:44PM -0600, Ron Johnson wrote: > > > On Thu, 2003-01-23 at 11:17, Pigeon wrote: > > > > Hi, > > > > > > > [snip] > > > > > > > > Having put appropriate deb-src lines in sources.list, pointing to CD > > > > images or websites as appropriate, have I got this sequence right? > > > > > > > > apt-get source --download-only whatever # get source package > > > > dpkg-source -x whatever.dsc # unpack it > > > > # fiddle with Makefiles etc. > > > > dpkg-buildpackage -b -ai686 whatever.dsc # create binary .deb > > > > dpkg -i whatever.deb # install it > > > > > > Here's how I do it: > > > # apt-get source gqview=1.0.2-1 > > > # export CC=gcc-3.2 -Wall -O2 -mcpu=pentiumpro > > Shouldn't that be -march=pentimpro, unless you want to generate > 386-compatible code?
Oh, man! Thanks for pointing that out... > > That's kinda nasty. I think CFLAGS and CXXFLAGS are better places to > > put this sort of thing. > > That's what I generally do, but some non-Debian stuff sometimes has > nasty makefiles that don't take account of your CFLAGS, or even your > CC, which leads to even uglier hacks like > > # cd /usr/bin > # mv gcc-2.95 realgcc-2.95 > # cat > gcc-2.95 > exec /usr/bin/realgcc-2.95 -Wall -O2 -march=i686 ${*} > ^D > # chmod a+x gcc-2.95 Blech! > or something along those lines. Yuk! (Raise both wings high above > back, whack offending code with one of them) > > gcc-2.95 v. 3.2: I've got 2.95 at the moment because that's what I've > been used to. Before I download big source packages, am I right in > thinking that 3.2 itself runs slower, but the code it produces runs > faster? And the difference is quite noticeable? So they say. On the few things I've tried it on, I haven't noticed. -- +---------------------------------------------------------------+ | Ron Johnson, Jr. mailto:[EMAIL PROTECTED] | | Jefferson, LA USA http://members.cox.net/ron.l.johnson | | | | "Fear the Penguin!!" | +---------------------------------------------------------------+ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]