Version 0.15.0 has been released, with major feature improvements. No more local patching with for sftp support for me! The gcc patch and gnome fixes in sitecopy-0.14.3.ebuild still apply and are necessary. So a simple cp sitecopy-0.1{4.3,5.0}.ebuild is all that is needed for version bump, and perhaps a glance at bug #87624 which is also still applicable to this release. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Hi guys (still need to get my @g.o buggzie account back so please pardon the strange address). Right now sitecopy latest is marked as follows: ~x86 ~amd64 ~ppc ~sparc (in about that order) I'm about ready to bump to version 0.15.1, but have added a couple of new changes to the ebuild, as we use neon unconditionally, and options that used to be based on system libraries are now based on what neon has built in (ssl, xml parsers, zlib). That said, I tested positive and negative combinations of neon USE flags with sitecopy, and it appears to be picking up just fine. That said, the ebuild here: http://dev.gentoo.org/~chriswhite/sitecopy-0.15.1.ebuild needs to be tested to be certain it works with your arches as well. Let me know of any issues you come across, and if it looks ok, let me know and I'll keep your keywords when I commit. Thanks ahead of time.
The behaviour of the new ebuild seems a little strange, at least it has suprised me. With USE including xml2 and expat, you receive an error saying "neon needs expat support disabled for xml2 support to be enabled". One would imagine disabling expat support for expat in neon will allow you to build sitecopy, but after neon is rebuilt with USE=-expat the sitecopy build fails with the "neon needs expat support" message. It would appear to me that the xml2 test shouldn't even run as neon with expat is apparently required, but I figure I must be missing the bigger picture. If I'm missing the bigger picture flames accepted, and a better error message for "use xml2" hoped for. (I would have suggested a better error string if I had actually figured out what was supposed to happen :/)
Well, if you build the actual source itself, you will notice (as I have), that you'll see a lot of "Using whatever neon uses" notices. Since we are using sitewide neon unconditionally, that's where sitecopy is going to look for things including expat. The actual neon expat/xml2 is kind of weird (expat disabled = xml2, expat enabled = expat...), so I thought that weird as well. Hope this makes sense (re-assigning too).
Yeah, my bad. What I meant to say, and now it isn't 6am I'm going to have a better try is... The error message for xml2 being set reads, to me, like the solution is removing expat from the neon build and not removing xml2 from the sitecopy build. I think really the "built_with_use neon expat" should say something about xml2 not working, as if you disable expat from neon then the previous test "built_with_use neon expat" fails before you check for xml2 in the sitecopy ebuild. In an example, in case I've not explained the point I'm trying to make well enough again: USE=-expat emerge neon; emerge sitecopy will fail because the sitecopy build doesn't fall back to checking for xml2 if neon doesn't have expat support, it just plain aborts on line 54. And yeah, all kinds of weirdness ;)
""" USE=-expat emerge neon; emerge sitecopy will fail because the sitecopy build doesn't fall back to checking for xml2 if neon doesn't have expat support, it just plain aborts on line 54. """ Well, you're messing with USE flags on command line, so I really can't support the theory you're giving me. This is basically why we don't like people doing USE flags on command line, it gives you undesired results and you should know what you're doing if you plan on that (key point). So in conclusion, you're giving a test situation that is wrong in itself, thus I can't support that theory (it seems that you have expat globally due to your situation?). Please correct me if I'm not explaining this correctly.
It was just supposed to be an example, so you could see how strange the error messages for USE with xml2 and expat set looked in context but as I'm the only one who has commented on this anyway it is probably a moot point. And as for global expat USE flag, yes it is set. Still, I added "net-misc/ sitecopy -xml2" to package.use and I've been happily using sitecopy 0.15.1 now since Friday so thanks.
Just a small note to the arch team. Weeve and myself found a pretty bad bug with the gnome frontend that wasn't hit on my side because I was using pre-existing configuration files, instead of creating new ones with the frontend. The gnome frontend happily segfaults after creating a new site definition and attempting to parse the xml. That said, I've disabled gnome support in the ebuild on my devspace until upstream can give me some status on support. If a fix does arise, I'll ask the arches to test it out later, just to be assured it works on all arches. Thanks for your patience!
Seems to work fine here on ppc, I downloaded a site and uploaded changed files with it. Anything that specifically needs to be tested?
Jose, I'm more concerned about the neon USE flag bit working ok. However, if you could test the sftp bit that'd be great. Toucan appears to use it for devspaces. All you really need to do would be to setup with toucan host/username/password and sftp as the protocol. That's it. Thanks again! Sparc, Please note that Weeve has ok'ed sitecopy being commited with ~sparc. amd64, r3pek has taken this task on for amd64 team. ppc, I'm sure you can tell that josejx is taking care of ppc side ;).
The following people tested this ebuild and reported it to me as a success: sparc: Weeve ppc: JoseJX amd64: r3pek I tested it for x86. Thanks for all the hard work guys :). In cvs and RESOLVED.
