On Fri, Oct 11, 2013 at 12:11:59AM -0400, Casey Daniels wrote: > > On 10/10/2013 11:35 PM, Ken Moffat wrote: > > On Thu, Oct 10, 2013 at 11:09:09PM -0400, Casey Daniels wrote: > >> Hello, > >> I'm trying to compile Squid 3.3.9 from Source and I keep getting the > >> error > >> > >> "checking size of size_t ... configure error in '/sources/squid-3.3.9' > >> > >> Configure error: cannot compute size of (size_t). > >> > >> here is the top of my config.log I have the whole thing, its just huge > >> and I also piped the output of the ./configure command if something in > >> it will be a clue. > >> > > What you probably want is the _last_ error in config.log. More > > specifically, look at config.log in 'vim' or 'less' (you are on > > ubuntu, so perhaps you don't have vim installed) and search for > > 'cannot compute size' i.e. > > > > /cannot\ compute\ size > > - type it just like that, hopefully there is only one such report, > > but if there is more than one hit '/' again to find the next one. > > > > My guess is that you are missing some of the infrastructure. > > Google thinks that autotools might be needed for this - perhaps > > 'apt-get install build-essential' will do the necessary, not sure. > > > > https://help.ubuntu.com/community/CompilingEasyHowTo > >> This file contains any messages produced by compilers while > >> running configure, to aid debugging if configure makes a mistake. > >> > >> It was created by Squid Web Proxy configure 3.3.9, which was > >> generated by GNU Autoconf 2.68. Invocation command line was > >> > >> $ ./configure > >> > >> ## --------- ## > >> ## Platform. ## > >> ## --------- ## > >> > >> hostname = build-pc > >> uname -m = x86_64 > >> uname -r = 3.8.0-31-generic > >> uname -s = Linux > >> uname -v = #46-Ubuntu SMP Tue Sep 10 20:03:44 UTC 2013 > >> > > Please note that we aren't set up to assist people running other > > distributions, so I'm assuming that you are trying this as a first > > step to see if you like building from source, and that you will come > > over to the light side in due course ;-) > > > > Also, ubuntu are still supporting 3.8 kernels, but 3.8.0-31 is well > > out of date, their latest is 3.8.13.11 ( I noticed the patches on > > lkml a few hours ago ). > > > > ĸen > > I'm actually trying to build it on a LFS box. I tried it in the chroot > environment. And it didn't work. (Thats what the Configure.log is from) > then I booted into LFS, it didn't work. I did however try it on the > Host system that I originally made LFS on and it did work. If at > possible I try to add packages to my LFS from the CHROOT environment > simply because I have access to web on the same computer and I can copy > and past the commands instead of typing them when following the BLFS > book. I love LFS and BLFS, just wish you guys had EVERY possible > package in there. > Too few people to build every package and to keep them up to date.
You do need to look for "cannot compute size" message(s) as I said originally, and concentrate on the last of them, because that is where configure decided your LFS system is broken. There are several things which can cause that breakage - among them are code which gcc now rejects, insane CFLAGS, and systems that really are broken (lost headers). It is the error messages from trying to run the code fragments which will point to the problem. None of the earlier error messages are interesting - all configure scripts check for system features by making fragmentary programs, like this one does at line 6451 - > configure:6451: checking if g++ supports C++0x features without > additional flags > configure:6488: g++ -c -g -O2 conftest.cpp >&5 > conftest.cpp:15:21: error: expected identifier before 'sizeof' > static_assert(sizeof(int) <= sizeof(T), "not big enough"); > ^ And that test can be expected to fail on some systems. > Here is the last little bit of the configure.log > > #define HAVE_STRUCT_TM_TM_GMTOFF 1 > #define HAVE_STRUCT_MALLINFO 1 > #define HAVE_STRUCT_RUSAGE 1 > #define HAVE_STRUCT_IPHDR_IP_HL 1 > #define SIZEOF_VOID_P 0 > #define SIZEOF_INT64_T 0 > #define SIZEOF_LONG 0 > > configure: exit 77 > Again, not interesting in the context of a failed configure - it is reporting how things have been set. These settings would determine how it will set up the Makefile(s). Also, which LFS version ? I see that Arch use the following: # gcc 4.6 doesn't support -fhuge-objects. sed '/^ HUGE_OBJECT_FLAG=/ s/"-fhuge-objects"//' -i configure that was pasted from https://projects.archlinux.org/svntogit/community.git/plain/trunk/PKGBUILD?h=packages/squid And the latest version is 3.4.0.2 - looks as if squid is like several other projects, with multiple supported versions. If your compiler is too new (that Arch sed implies this) then the newer series might be easier. OTOH, fedora are using 3.3.9 and their most likely change is to patch src/ssl/support.cc so it obviously configures for them, albeit with a lot of other patches. http://pkgs.fedoraproject.org/cgit/squid.git/tree/ ĸen -- das eine Mal als Tragödie, dieses Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
