"Alexandria is a GNOME application to help you manage your book collection."
The program also depends on ruby-amazon which is not in the tree, so I created an Ebuild for that too:
http://bugs.gentoo.org/show_bug.cgi?id=47941 (ruby bindings for GeoIP which can be used by ruby-amazon)
Created attachment 29354 [details]
Merges and unmerges fine on ~x86. As I wasn't sure which category it belongs
to, I put it into apps-misc in my overlay.
gnome peoples - interested?
nope, afaik this pack is early in development anyway so shouldn't be added at this time.
the ebuild misses the header
there is no 'GPL2' license
if it is a real gnome app it should probably use the gnome2 eclass
the config step is done twice (?)
dunno much about ruby, but shouldn't those deps be RDEPs ?
foser, you are right, it is early in development. I just wanted to provide an ebuild for people who want to try out the package.
Sorry for deleting the header, my fault. As well as the license, which should read "GPL-2". For the double configure step: I don't know too much about Ruby either, that's what I got from looking at similar ebuilds. And you're right about the RDEPENDS I guess.
To sum it up: I pretty much screwed up on the ebuild, sorry. But as you mentioned the app isn't production ready anyway, I just wanted to make it accessible to people willing to test it. I hope that at least the other 2 ebuilds I did in the process are worth something and will hit the tree eventually...
Thanks for your comments though!
The Ebuild can be bumped to 0.1.2, which is a little more useful (e.g. working ;) than the version before.
Yeah, there's probably no one interested besides me, it's just for the record...
foser, thank you for your comments!
I did a new ebuild for 0.1.2, not because I think it is that important that the package hits the tree, but more to improve my skills at making ebuilds.
1. Missing headers added
2. License corrected to GPL-2
3. Dependencies corrected to RDEPEND
Regarding the double configure step: checked several ebuilds in the dev-ruby category (e.g. fxruby, madelaine), it really seems that's the way to do it if install.rb is used.
As a result of this repoman now is happy with the ebuild, besides a missing ChangeLog and metadata.xml.
Created attachment 32196 [details]
Corrected ebuild for Alexandria, not just a version bump
Merges and unmerges fine on a ~x86 system. Works ok so far taking into account
the age of the program.
Your ebuild looks tidy and clean ;) However, it needs some more work.
Dependency to ruby-libglade2 is missing (without it, alexandria won't
work). DEPEND list should consist of so-called DEPEND Atoms (see
ebuild manpage). Bare package names are not allowed. Please reattach
your ebuild when you finish fixing them and I'll review it again
and commit to CVS. Thanks in advance.
Ok, you're right about ruby-libglade2. I initially believed it to be a part of ruby-gnome2, which it apparently isn't. ;) Problem was that I had it installed, so I didn't notice the problem.
But I don't get what you're telling me about the depend atoms, because mine look like
They aren't just bare package name. Or are you talking about something else?
Anyway, I think I won't fix this version, because 0.2 is already out. I have an ebuild for it which makes proper (I hope) use of the gnome2 and ruby eclasses, but unfortunenately I get some errors when installing GCONF schemas. I'll have a look into it and then submit hopefully submit a working ebuild for 0.2.
In short, you need to include package categories for each package.
e.g. >=dev-ruby/ruby-gnome2-0.9 (you omitted 'dev-ruby/')
See, for example, Common Gentoo Ebuild Mistakes
I'm _so_ embarrased for not seeing the obvious... This whole Alexandria ebuild seems to get over me anyways (see the email I sent you). As you may have noticed I didn't do this mistake in any of the other ebuilds I submitted, seems I somehow messed up just this one.
Anyway, as this currently is going nowhere, I'll just stop making noise until I have figured it out a little more...
I've finally added Alexandria 0.4.0 to CVS, this bug can finally be closed.