I'm trying to come up with a set of patches that will go on top of the
vanilla 2.6.7 kernel sources, and have all the S/390-specific stuff be
current.  Before I start complaining (and I will get there), I want to say
that the current state of affairs is _much_ better with 2.6, than it is with
2.4.  Most of the patches that IBM Boeblingen have published for 2.6.5 are
already included in 2.6.6, and 2.6.7.  That's really nice to see, since it
means I won't have to be looking after tons and tons of patches as time goes
on.  But, (now the complaining...)

What I did was take the 2.6.5 kernel and put on all the developerWorks
patches to bring it up as current as it could be.  Then, I ran a diff
against the updated 2.6.5 source tree, and a vanilla 2.6.7 tree (comparing
only the files touched by the developerWorks packages).  As I said above, a
large number of updates were already incorporated into 2.6.6/7, which was
nice.  What was not so nice is the fact that _some_ of the S/390 modules in
2.6.7 had higher version numbers than the supposedly current 2.6.5 modules.
Also, some of the updated 2.6.5 modules had higher version numbers than
2.6.7, but that didn't surprise me much.  EXCEPT, some of those updates were
in the linux-2.6.5-s390-base-april2004.diff file, and so should have been in
2.6.6 and 2.6.7.  Argh.

I've run into difficulties with other packages that IBM modifies, but was
usually able to figure things out on my own.  The reason for that was
because the current source code was available via CVS or a web interface to
CVS.  I could see for myself what the developers thought the current state
of a particular module should be, and figure out how to get there from where
ever I was starting.  I cannot do that with the Linux kernel code, since the
CVS/BitKeeper repositories for the S/390 developers are not publicly
available (to my knowledge).

Is that ever going to change?  If so, when?  If not, why not?  One of the
good things about the Open Source development model is the transparency
involved.  I'm once again feeling like a mushroom when it comes to the
Linux/390 kernel source.  Some more transparency would really help out right
about now.  If nothing else, a daily snapshot of the S/390 portion of the
tree would be a nice start.


Mark Post

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to