Randy McMurchy([EMAIL PROTECTED])@Fri, Aug 12, 2005 at 08:09:44PM -0500: > Randy McMurchy wrote these words on 08/12/05 17:11 CST: > > Ag Hatzim wrote these words on 08/10/05 04:15 CST: > > > >>Log of the already installed packages attached,most of them confirmed to > >>work > >> > >>libxml2-2.6.20 > > > > This is odd. LibXML2 crapped out for me during the 'make'. > > > > runtest.o: In function `xpathDocTest': > > /home/rml/build/libxml2-2.6.20/runtest.c:2404: undefined reference to > > `rk_glob' > > /home/rml/build/libxml2-2.6.20/runtest.c:2412: undefined reference to > > `rk_globfree' > > This must be a Glibc issue. I'd bet anything Ag is using a Glibc > snapshot and not the latest stable version. >
Too bad for me i wasn't here around to bet with you :-) No,my build was based on the cvn of 30 of July with some cosmetic changes by me but nothing serious to affect the glibc built. I already saw that your problem started from Heimdal was overwritten the glob.h which installed by glibc,but that made me think and probably (still not sure though) change my thought in a previous post . Maybe a gcc4 BLFS branch as you suggested in your first post,Randy? And i explain. My way to build blfs packages,is through scripts and functions. The first variable however that my scripts read is the gcc version variable (thanks to Greg), so depending from the gcc version they follow a different profile (different instructions if any,different patches,different folders,etc ...). I am not really sure if a note in the blfs instructions e.g "If you build with gcc4,follow that...or patch it with this gcc4 patch etc..." is enough. I know that is not a rocket science and if (I) can follow the instructions ( i am using Linux in the last 15 months) everyone can follow them i believe,but with my poor knowledge i am considering gcc4 as an entirely different compiler. I know for sure there are still many packages (some are not in the BLFS),that even they are builted successfully (not to mention the uncountable warnings during the compiling) still refuse to work properly,and 2-3 of them fail miserably. There are now,enough people to test the 4* gcc LFS branch,but its too early to say "It is working as it should be"...yes i know there is a contradiction with a previous statement. As proved by libxml2-Heimdal-glibc,problems will occur with different combinations,and special notes under notes,even in BLFS development stage book,they are enough to confuse people and maybe (maybe not) will slow down the development. BLFS must be ready and follow,when LFS gcc4 go stable,with instructions for how to build BLFS depending *completely* in gcc4 LFS and for sure not again with notes,such as "But if your compiler is gcc 3.4* there is no need for this patch ...etc." These are my thoughts,and this is how i see things from a common user point of view,thus you should considering them like such ( i really don't know the amount of the work that needed for 2 different books depending by the compiler version). Personally i 'll happy with the one way or the other,i am happy and proud with a mysterious way,that i am in L(B)FS neighborhood (almost 10 months now), and i can only be anxious for the future. For example i will be really happy to see LFS in the near future being a a platform where you can treat it as your base working system,and go away from the common (false) motto floating around. "LFS is just for learning Linux,but maintenance is a PITA,so i am working in debian-ubuntu-fedora etc..." But that is an entirely different conversation,so keep it for the rainy days... that time of season is still quite hot for philosophical discussions :-) With Regards and sorry for the awful english. -- http://linuxfromscratch.org/mailman/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page