Skip the "tests" in the build can cause problem (in chapter 6)? [doubt]
 

Folks, I want to know if skip those "tests" before compilation can cause some mistake, I already know it will give error in these tests in some packages.

For example, before compiling the GCC I do the tests: make -k check. It takes hours to complete, is it okay to skip it?

Well, skipping the tests can't cause a mistake. But if something is wrong, you won't know about it.

Personally, I do the tests that the book says are critical. But I never do any of the other test suites, and I never do any test suites for BLFS packages. You're right...they take too long. Furthermore, I have never had any test results that were bad or unexpected. Since I build LFS and BLFS with scripts nowadays, I tee those test results to text files and look at them later. One test suite in particular, GCC, will stop a script. For that one test suite, I backup the build directories and run that long test suite after the scripts have finished. Anyway, some people here don't run the test suites. Here, for example, is a post where Krejzi gives his opinion on not running test suites...

http://www.linuxquestions.org/questi...0/#post5451050

I have always considered Krejzi to be Armin K. who is/was one of the LFS developers. Anyway, he sure seems to know what he's talking about around here.

Stoat, thanks for answering, I had already read this part of the book before actually had completed, but I inadvertently deleted the partition and lost files, then wanted to spare time to redo it. As I had done before (before deleting the partition, kkkkk) so I guess I'll have no problem with any other unexpected errors that may show up in tests, I believe it will not appear any other error.

I do the same as stoat. I run the tests that the book says are critical. Most other tests I ignore unless I'm experimenting with optimization or settings and I want to know if things are going well.

"Ciritical" tests are useless. Most of them only check how well a package conforms to a latest standard (who cares who builds LFS anyways?) and if a bug has appeared again during the development cycle. Sanity checks are good enough to see if something went wrong. If they are not according to the book, you screwed up. And test suite won't detect it, because it will pass - you'll just get a system that links to /tools - again: those tests are useless.

I only do the the checks to make sure that stuff is compiled correctly and the right lib path and include path is searched etc and only in chapter 6 anyway, never yet had any problems, touch wood.

As I said, make check won't reveal that. It will pass, even if the toolchain links to libraries in /tools. Sanity checks are the only way to find that out. Rest of the gcc/glibc testsuite is useless.