On Sat, Mar 25, 2006 at 03:28:20PM +0100, Andi Kleen wrote: > With CC="ccache cc" it works as expected. And here I was, thinking that CC should be a variable containing no whitespace, ie. just a single path to a binary, be it relative or absolute, and that there were ways of teaching ccache to be completely transparent (using gentoo, it was set up for me that way): $ ls -l /usr/lib/ccache/bin/ insgesamt 0 lrwxrwxrwx 1 root root 15 6. Mär 01:13 c++ -> /usr/bin/ccache lrwxrwxrwx 1 root root 15 6. Mär 01:13 cc -> /usr/bin/ccache lrwxrwxrwx 1 root root 15 6. Mär 01:13 g++ -> /usr/bin/ccache lrwxrwxrwx 1 root root 15 6. Mär 01:13 gcc -> /usr/bin/ccache lrwxrwxrwx 1 root root 15 6. Mär 01:13 i686-pc-linux-gnu-c++ -> /usr/bin/ccache lrwxrwxrwx 1 root root 15 6. Mär 01:13 i686-pc-linux-gnu-g++ -> /usr/bin/ccache lrwxrwxrwx 1 root root 15 6. Mär 01:13 i686-pc-linux-gnu-gcc -> /usr/bin/ccache $ So all I need to do in order to enable ccache is add this stub dir to the front of my PATH: $ echo $PATH /usr/lib/ccache/bin:/usr/lib/distcc/bin:/usr/local/bin:/usr/i686-pc-linux-gnu/gcc-bin/4.1.0:[...] (Note: the order of distcc after ccache is important, if not obvious) So what the thing does is call the next instance of the name it was called by from your path, ie. subtracts dirname($0) from the beginning of $PATH, and supply each and every argument to that. In my past setup, stuff is taken from the cache, and if not found, distributed across machines. Works extremely well for kernel builds on slow machines, with a faster one right beside it. And no dirty CC= stuff to begin with : ) AFAICT this is not a religious or purist issues, as there have been real problems with CC not following the recommendation and some projects relying on that, but as this thread goes to show there occur problems even without real problems existing ; ) Maybe setting obscure CC variables works reliably for kernel builds, I never tried... Kind regards, Chris
Attachment:
pgpnAnms8DM4I.pgp
Description: PGP signature
- References:
- ccache complains in latest git about -p
- From: Andi Kleen <[email protected]>
- Re: ccache complains in latest git about -p
- From: Andi Kleen <[email protected]>
- ccache complains in latest git about -p
- Prev by Date: Re: Help related to socket creation in kernel space
- Next by Date: Re: [2.6 patch] usb/input/keyspan_remote.c: don't use an uninitialized variable
- Previous by thread: Re: ccache complains in latest git about -p
- Next by thread: Re: ccache complains in latest git about -p
- Index(es):