begin  quote
On Wed, 24 Dec 2003 14:17:04 +0100
Christian Gut <[EMAIL PROTECTED]> wrote:

> 
> Ok, i can understand that, but what about masking it for ~arch or ~x86
> like those other things (glibc)?

okay, I come into this discussion from a QA perspective:

THAT IS NOT AN OPTION!
*weilds cluebat*

introducing -known-to-break- changes to ~x86 is once more
NOT AN OPTION.

~x86 is not "developers playground where things should break and you're
on your own if things dysfunct"

~x86 is -EBUILD- testing stage, as in the -software- and its known
subdependencies -SHOULD- work, and unless you have a darn good reason
(this isn't that. sorry, back to go)  for introducing known-bork to ~x86
I shall have that persons head, ebuilds and other things slewed out to
the lists and net.  package.mask exists for this reason. 


Now, That I've had my fit,
   until you have a working upgrade path from vanilla 2.4.23 to 2.6
(Default config that enabled devfs is a nice idea; )  And all (Repeat)
all sub-builds in the tree are working, or have fixed dependencies that
they only can live on a system with 2.4 kernels,  moving 2.6 to vanilla
isn't an option.


Yes, that means that somone has to try 2.6 on about three or four
thousand packages.  Albeit most of them are pure userspace, debuggers
(Valgrind somone?)  and kernel modules and such interfaces much be
redocumented to work.

Important change for example: anyone relying on pcmcia-cs -drivers- (I
was, the cardbridge in 2.4 didnt reset properly at times)  Will have to
have a documented upgrade path, or they will be in the icy shallows with
a broken system post-boot.


So, that concludes my rant.  Don't consider ~x86 to be your local broken
playground. 

//Spider

-- 
begin  .signature
This is a .signature virus! Please copy me into your .signature!
See Microsoft KB Article Q265230 for more information.
end

Attachment: pgp00000.pgp
Description: PGP signature

Reply via email to