On 02/12/10 10:36, Shawn Walker wrote:
On 02/11/10 10:11 PM, Bart Smaalders wrote:
On 02/11/10 12:19, Shawn Walker wrote:
...
Overall, this seems right; sorry for the incomprehensibility of
t_api_list.py.
However, I did have one concern -- will verify/fix handle this situation
correctly? I've looked through the repair code, but it isn't clear to
me what will happen.
I've updated the code to handle chained renames, and now uninstall all
possible renames whenever a package is uninstalled.
pkg verify & fix don't do this yet...
However, add or remove a package or image-update w/ these bits, and
we handle things.
This ok? I'll wait w/ pkg fix until next release; I want to
go over that in considerable detail to handle more egregious
damage that it can cope w/ at this point.
modules/client/imageplan.py:
line 237: Do you really need access to all known packages for the
solve_uninstall case, or just the installed packages? If you only need
access to the installed packages, then use
self.image.IMG_CATALOG_INSTALLED here instead. The INSTALLED image
catalog is identical to the KNOWN catalog except that it only contains
the installed package data; the state information, etc. is all still
there. That should reduce memory usage and improve performance of this
operation.
I'm going to leave this the way it is for now. I anticipate that some
of the new dependencies we're considering may result in pkg
installations as a side effect of removing other packages, just as this
push will result in removals as a side effect of installation.
...sorry for being a bit slow on the response. The solver code makes my
head spin a bit :)
Some days it has the same effect on me ;-)
Thanks for the review.
- Bart
--
Bart Smaalders Solaris Kernel Performance
[email protected] http://blogs.sun.com/barts
"You will contribute more with mercurial than with thunderbird."
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
