If you read my message, you'll note I said I wanted to make sure it was
compilable with different configurations. What I'm talking about only entails
compiling the kernel multiple times with different config.in files. Where did
you get the impression I wanted Linus himself to test the drivers thoroughly
himself? I only was suggesting he should find some automated way to compile
the kernel with multiple configurations and look for typo-type bugs.
If Linus only did this with a reasonably chosen subset of the possible
configurations, then the only costs to him are the few hours for each release
this test would take, the space to compile a kernel, the CPU time for these
compiles, assuming his computer is on 24x7 anyway. I think this is a
reasonable suggestion.
> It offends me that it sounds like you aren't satisfied to get a chance to
> help rather than trying to be a straw boss. Do you think Linus intends to
> release things with typo's. Haven't you ever made a typo in a final draft.
> I was looking over my Published MS thesis the other day and still found
> typo's And I certainly wouldn't expect Linus to go thru that hell for every
> release. What did it cost you to correct the typo?
I'm making a suggestion. I realize Linus probably doesn't intend to release
typo's but I don't see much more than cursory attention being paid to them
before the release (I don't mean to be unfair here - if Linus does test
compiling the kernel with many config options, this entire discussion is
useless).
Typos in the kernel cost me the time it takes to track down the fix on my own
or the time I spend reading the newsgroup to find the fix. The thing is,
anyone who wants to compile the kernel under the problematic conditions has to
spend this amount of time as well. If the kernels are released another 4-6
hours later than they are ready but are free of typos, I think on a global
scale you save time. The important thing to note is this lost time is
avoidable, and I'm suggesting that we try to avoid it.