On 04/28/10 09:03 AM, Glenn Lagasse wrote:
Hi Keith,

* Keith Mitchell ([email protected]) wrote:
Hi all,

I'd like to request a review of the following changes to DC and AI
default manifests that change the packages listed from the old
"SUNW" names to the current hierarchical pkg names:
http://cr.opensolaris.org/~kemitche/pkg_rename
Some of my comments others have mentioned:

1) # of packages don't match between the old and new manifests.

In the AI SPARC case, the SUNWpd package was included, but that is an x86-only driver (driver/storage/glm). The glm driver for SPARC is included in driver/sparc/platform, which was already in the list.

For the text installer case, there were some mismatches in content between the x86 and SPARC lists, and I wanted to address those differences while I was working on the manifests. To track this properly, I filed bug 15786 and will update the commit message to reference that bug as well.

2) You removed the comment about the ordering for entire, SUNWcs and
SUNWcsd which I understand since pkg fixes that up internally now.
However, I wonder if we shouldn't have some sort of comment that those
three packages are *required* to be in the manifests (since they are).
My thinking is to avoid 'customizers' coming along and removing them in
the name of generating a minimized image and then seeing it not work.  I
know the original comment didn't really state this, but I think it did
sort of imply it.  This could be a slippery slope, since you need a lot
more than those three packages to get a useful image for any of the
image types, but I think those three packages are at least the minimum
requirements to get started as it were and perhaps deserve a special
mention.  What do you think?

I sort of lean towards the slippery slope justification. There's a desire to define a "minimal install" set of packages; once that's available, including that meta-package as a base, and indicating that it's required seems to make more sense. The im-pop stage itself will at least fail to install the packages if SUNWcs and SUNWcsd is missing, so it will be obvious that those are needed. "entire" is a bit more ambiguous, though the remaining comment about appending a build number to "entire" carries the same pseudo-implication of requirement as the old one did.

Otherwise, looks good.


Thanks for reviewing!

- Keith

_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss

Reply via email to