18 Jan
2011
18 Jan
'11
8:06 p.m.
On 18 January 2011 18:51, Pierre Schweitzer
<pierre.schweitzer(a)reactos.org> wrote:
Why is that an "old" point? We've only recently started discussing it.
Hi,
For my information??
You need to start your sentences a little more politely
Whilst parts of our kernel try to target 5.2
changes, from a testing
perspective it should be
pretty much identical to 5.1.
IMO, unless you're a kernel dev interested in implementing some of the more
elegant kernel
features in 5.2, then anyone with a 5.1 box is more than adequately setup
and I can't think of
any worthwhile advantage in having a 5.2 install.
For your information, between NT5.1 and NT5.2, MCBs handling has been
rewritten. This is not a minor
change, as it brings the idea of "base" MCB for dealing with large MCBs. Then
testing against Windows XP looks just really hazardous for such case.
We are targeting Windows 2003 for kernel, why the hell would we test against something
else?!
I started my my point with :
"unless you're a kernel dev interested in implementing some of the
more elegant kernel features in 5.2"
of which you and MCB's fit this statement.
You're talking about testing mappings via the FsRtl kernel API's,
which isn't important to an end user nor 99% of devs who work on
reactos.
It's a feature you choose to implement with regard to how the control
blocks are handled.
Would anyone actually use a remote machine for
development?
I'd expect most devs to already have windows machines, I can't see how can
you be a ros/nt dev without access to at least one NT5 box.
Then, to answer that "old" point. I've a Windows devenv available (MSVC, and such)
but on Windows 7.
Which is no use for such tests. I also have a Windows 2000 available,
but once more... It's NT5, so...
I keep pushing, it IS useful for the project.
You mean it's useful for you.
Furthermore, how could we explain we implement Windows
2003 without having any Windows 2003 for our devs?
Most reactos devs aren't kernel devs so that a null argument as we
implement the Windows 7 API set in usermode.
Secondly, anyone who states that we implement a 2k3 kernel is wrong.
Alex used to target 2k3 in the days of his kernel rewrites
but the fact is that most of our kernel is very much NT5.1 based (due
to the similarities) and in many cases totally different than any
of the NT kernels.
I'm not sure why you felt the need to be rude/aggressive, perhaps you
feel threatened in by my question which now appears to be
"Would anyone other than Pierre have any use for a (free) remotely
available copy of 2k3?"
If you really need it, you can probably make a good case to the
foundation for your own copy.
My point is, is it really worth having this online server just for
your use when it would be better served as an MSVC build slave which
would benefit everyone.
Ged