Verification of Build Procedure

Christos Gioran himicos at freemail.gr
Sat Nov 29 03:55:18 PST 2003


On Saturday 29 November 2003 05:26, Greg Schafer wrote:
> > I was wondering whether there is a way to verify that a LFS system was
> > built correctly.
>
> Umm, the best way it to just *use* the thing and see what breaks :-)
>
> > pass, there rises the need to detect overoptimization. Are the testsuites
> > designed for this purpose?
>
> overoptimization ?? Test suites check functionality - optimization does not
> come into it at all. If code is optimized and still functions correctly
> then it will pass its test suite - simple as that.
>
> > So to conclude: Is there a way to thoroughly verify the stability of a
> > build, or should I (and all who might want want, of course!) come up with
> > one?
>
> Your statement of requirement is too general. Of course there are various
> 3rd party test suites and benchmarks around the place, but they mainly
> check a small well defined part of the system. How do you define "thorough
> stability of a build" ??
>
> You could always do what the distros do, make a distro, ship it to
> thousands of unsuspecting beta testers, see what breaks :-)

Well, by stability I primarily mean run-time errors that occure due to 
optimization flags. Now that I think it over, I realise that it is not 
possible to test ALL possible usage scenarios with a test suite. The beta 
testing guinea-pigs are a good (if not the best) solution. Probably it is 
best not to optimize when building production systems

Thanks for your time

himicos



More information about the lfs-support mailing list