Simon Geard wrote: > On Fri, 2011-06-17 at 11:12 -0500, Mike McCarty wrote: >> Webmaster wrote: >>> I never "check", because if the check passed it's useless but if the >>> check failed you can do nothing. >> Then you do not understand the purpose of testing. I've heard >> many a manager say more or less the same thing. > > Harsh. Automated tests are primarily for developers to spot regressions > in their own code - if they happen to be useful to spot problems in an > LFS build, that's just a bonus for us. > > Personally, I don't bother - they're critical when I'm writing code, but > I rarely take the time to run them when simply installing someone else's > work.
I always run the tests in Chapter 6 when installing a new package version. If I've installed a package before and it was OK, then I don't feel the need to test again. I like to understand any problems before using a package. For example, bc gives about 10 errors out of 1000 but the nature of the errors are round off issues that I can live with. Some of the errors identified in LFS packages are quite tricky. Timing errors are especially difficult, but there have been cases where there were bugs and we held off a package until we got a fix. GRUB-1.99 is a case in point right now. I still have to try a patch to see if I can get it to work properly with LFS. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
