At Sun, 20 Feb 2005 14:49:16 +0100 Holly Bostick <[EMAIL PROTECTED]> wrote:
Dan Johansson wrote:
OO.o is just "weird" (for want of a better word); the difficulty of compiling it, in addition to the length of compile, is the reason the binary is available in Portage. I suspect that the difficulty of compiling it is the issue in this case, because it seems quite likely to me that one of the "weirdnesses" could well be binary components squirreled away in that extremely long compile, which revdep-rebuild doesn't know how to (or is not built to) deal with.
So both of these programs likely share a common weirdness factor of having some binary components alongside some compiled components (again, IANAP, this is my impression from working with them in the past).
I can't comment on any weirdness in OO.o or confirm/deny that it has binary components. But I can say that just having OO.o emerged (from sources) does not by itself cause revdep-rebuild to complain. I have OO.o and just now (right after a sync) tried revdep-rebuild and it found nothing to do.
I understand that, but on the other hand, nothing is apparently broken on your system requiring a rebuild in the first place. In my experience, if a rebuild has already been performed to fix actually broken stuff (this may be the defining factor), and OO.o is present (usually as a binary, but I have also compiled it from source, but I use OO.o-ximian, so maybe that's it as well), OO.o will always come up on the "check" rebuild (the second one you do to make sure that everything got done on the first one). Since it was already present in the first, "real" rebuild, I don't do it again, and consider the first rebuild successful and complete (as OO.o works fine anyway and was never apparently truly "broken" at all-- again, in my experience, and I don't use OO.o that heavily, so it might have been broken in a way I wasn't aware of). But after that first rebuild, the only thing that actually appears in the list of potential rebuilds is OO.o (I don't have opera, either), and nothing else, ad infinitum (until I break something else, at least).
I wouldn't expect it to come up if a revdep-rebuild was run and found nothing otherwise wrong (that would be a real bug). I didn't mean to imply that I did, sorry.
I always pay attention to posts involving revdep-rebuild, since I depend <groan> on it working correctly.
[ snip ]
allan
PS Holly, I also often read your posts as I find them well-written and helpful.
Thank you very much.
Holly
-- [email protected] mailing list
