Hi Jonathan. I compared the preprocessor output between pass and failure cases, and the diff I saw gave me some hint that there were few headers in the new installation. I never knew that only those headers that should be fixed were copied under fixed-includes. We then checked our code, and realized we had to make some changes to get the compilation going on redhat6.x So, yes, it is not really a problem with the fixedincludes, instead our code. On Fri, Feb 21, 2014 at 1:19 PM, Jonathan Wakely <jwakely.gcc@xxxxxxxxx> wrote: > On 14 February 2014 02:50, vkr wrote: >> I was building gcc 4.4.3 on Red Hat Enterprise Linux Server release >> 6.3 (Santiago). Build has passed, however, fixedincludes directory is >> missing some headers which were present in the redhat5.5 installation. > > Why do you think that's a problem? > > Only headers that need to be fixed will get fixed. If the headers on > rhel5 and rhel6 are different (which of course they are) then a > different set of headers will get fixed. > >> It seems to me that because features.h is picked from the system >> header on Redhat6.x, when I compile my project source tree, not from >> the fixedincludes, as it is missing anyway, compilation is failing. > > I don't understand this paragraph. > > Are you saying your project doesn't compile? > > Why do you think that is related to fixincludes?
