[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Condor-devel] The Great Renaming(tm) of "vm" => "slot" now committed to HEAD
- Date: Wed, 30 May 2007 15:20:06 -0700
- From: Derek Wright <wright@xxxxxxxxxxx>
- Subject: [Condor-devel] The Great Renaming(tm) of "vm" => "slot" now committed to HEAD
hello world,
i just committed the giant patch to our source to completely purge
the "vm" and "virtual machine" terminology from condor, except for
Xen/VMware VMs, Java VMs, and PVM. ;)
the doc changes are incomplete, but i was tired of maintaining the
patch, resolving conflicts, etc, while i was trying to find time to
finish the docs among all the other fires i'm fighting right now. my
trusty regexp finds 113 more lines of the manual i should visit and
edit before i'm completely done. but, at least the code is now in
the repository (and therefore, will make it into 6.9.3), and the bulk
of the manual changes are in place, too.
a few important points:
1) there's a bunch of backwards compatibility code sprinkled
throughout the daemons and tools to deal with the old config
settings, classad attributes, command-line args, etc. in other
words, this change is fully backwards compatible. however, if you
don't want the additional classad attributes, and want to move to the
glorious new world order, you can set the following knob:
ALLOW_VM_CRUFT = FALSE
in your config file, and none of the backwards-compatibility code
paths will be hit. for 6.9.x and probably 6.10.x, this knob will
default to TRUE. for 6.11.x, we'll probably default it to FALSE, but
leave it there for people who *really* need to maintain compatibility
with older condors. by 7.0 (or whatever comes next), we'll hopefully
just prune all the code of these crufty code paths completely and say
"7.0.x doesn't talk to 6.8.x, get on with your life already".
2) as is our stated policy, we always assume everything on the same
host is running the same version of condor. there are a few places
for intra-machine communication that have no ALLOW_VM_CRUFT code
paths, and simply won't work if you've got a 6.9.3 daemon talking to
an older daemon on the same machine.
3) if anyone commits any changes to HEAD that re-introduce "vm" or
"virtual machine" terminology where you really are talking about
"slots", either in log messages, variable/function names, classad
attributes, etc, you will face my wrath. ;) that also goes for
merges from V6_8 -> HEAD, so if you're doing such a merge, please
beware.
stay tuned for another update about the completion of the doc
changes, hopefully coming soon.
enjoy,
-derek