Danek Duvall wrote:
Shawn Walker wrote:
http://cr.opensolaris.org/~swalker/pkg-11384/
in add_publisher(): why remove the calls to remove_publisher_metadata()?
Don't you want to remove the data after a failure? Should these be changed
to another call to remove_meta_root()?
No. My belief is that I erroneously moved these there. The original
code did a single remove_meta_root() before calling refresh_publishers.
My belief is that refresh_publishers() should be able to handle all of
this automatically so I don't need to do anything.
line 1354: this should use whatever constant you've introduced for bug
11735
Rich is working that bug, not me. But since it hasn't been putback yet,
I can't do that :)
I'd probably put the two calls to rmtree() (to remove the old catalogs)
after both renames have been done. You might also want to use a new
temporary name instead of appending .orig, since if something happens and
those .orig directories don't get removed properly, the rename will fail.
Though better yet, you could just create a temporary "state" directory one
level up, and reduce your window even further.
I'll take that approach.
Should temporary_file() get the same try/except bits that temporary_dir()
has?
Yes; that was just tunnel-vision on my part.
Cheers,
--
Shawn Walker
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
