lkml.org 
[lkml]   [1997]   [Jan]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRE: too much untested code in new kernels
> Hello hackers,
>
> I see a lot of new code in recent kernels which has not been tested
> first.

Hence my request for Linus (or _somebody) to make a test compile with every
possible option turned on, to make sure the d*mn thing will COMPILE at least!

Whenever anybody mentions this, they get shouted down, claiming (validly)
that it is impossible for anybody to test all possible combinations of
options.
But for CRYING OUT LOUD how about just making sure it compiles!!!!!!!! And
verifying that the resultant image boots up.

Even better would be for Linus to refrain from making a release (or
accepting patches) unless it compiles "clean 0/0" (zero warnings and zero
errors).

This shouldn't take a lot of extra time. My 6x86/p166+ only takes 8 minutes
to do a compile. As I recall, Linus has an Alpha that's even faster. Then
add another few seconds to grep the log, and check for warning/error
messages.

Ray






>
> The top-level MAINTAINERS file says:
>
> 1. Always _test_ your changes ...
>
> 2. Try to release a few ALPHA test versions to the net. Announce
> them
> on the kernel channel and await results.
>
> The new module code was not tested in this manner.
>
> The watchdog driver has not been tested in this manner. In fact,
> drivers/char/pcwd.c has not even compiled since 2.1.15! This is not a
> subtle CARD_REV configuration-dependent bug, and it's not a case of some
> other kernel interface changing and this driver needing to catch up. It
> is an outright mismatched open brace in the middle of pcwatchdog_init!
>
> When I see things like this, it makes me very hesitant to run new
> kernels. I see 1400 of lines of changes in the 2.1.18 VFS and EXT2
> code. Have these changes been tested? I don't know! Are they getting
> tested right now? Not very much, because everyone who uses modules is
> struggling with the incomplete module code that was also introduced in
> 2.1.18.
>
> So the results is that I'm not going to run 2.1.18 or any later kernel
> on my machine for several weeks.
>
> Please, code authors, test your patches before you send them to Linus.
> And please, Linus, if a patch has not been tested to your satisfaction,
> tell the author to test it some more before you accept it.
>
> Michael Chastain mec@shout.net "love without fear"
>

\
 
 \ /
  Last update: 2005-03-22 13:38    [W:0.036 / U:0.220 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site