This is a quick fix for the BZ.
There are no cucumber tests included, since there are no steps already defined for images at all, other than upload XML and template.
Implementing the full list of steps to get to the point where we can test this small change is not practical in this commit + we have a deadline coming up.
There are a lot of issues in the Images Controller and View as well as in aeolus-image builders model. These are out of the scope of this patch. Next iteration I propose we do some serious tidying up these areas.
To test this patch.
You need to kick off a build that takes some time. (I've tested with RHEVM). Then go to the show page of the image you have just created. There should be no build button, in its place should be the state of the active build.
This is a resend of my earlier patch, "BZ 795743 - Detect architecture for imported images" in response to Jan's feedback. Namely, he pointed out that, since we're storing this architecture attribute for imported images, we might as well *display* it. I added that in this patch.
I got sick of writing "image.os.arch || image.architecture", so I added an "architecture" method to images in aeolus-image-rubygem to do that for us. This patch is already introducing changes to aeolus-image-rubygem so this seemed tolerable, though this change does mean that you'll get exceptions if you upgrade Conductor and not aeolus-image-rubygem. (So I'll send out an [ANNOUNCE] message noting this after this patch is ACKed but before pushing, as someone proposed earlier today.)
Just noticed a type in "man aeolus":
"This contains the URL and credentials for you amdin user in aeolus-conductor"
"admin" is spelled wrong.
Anyone want to push a trivial fix for it? (not worth a bz?)
Aeolus Community Manager
Fixes a view that I missed, and resets the cuke tests - apparently they
were not failing becuase cucumber wasn't rendering t() strings as I thought,
they were reporting a legitimate error! Who'd a thunk it...
I'll save it for tomorrow to do a proper writeup and finish the mockup
but i've added some of the things i've been working on
(with lots of help from Tomas) here.
I'll merge it into the conductor UI repo soon too
Basically this gives more power to the stylesheet and simplifies the html view structure and linking workflows hugely.
a very immature state of an example app is here. no it won't work in old browsers but now supporting them will be much much easier.
(sorry about the internal link)
Four things to think of about
* any view can be activated or have any content from it inlined simply with a link.
* no stylistic classes to worry about remembering, functional classes only (e.g. .active not .tab or .left-120)
* no bitmaps or images to maintain, slice or cringe at when scaled
* Eventually three http requests only (css,js,html) (which we could even inline to one)
* freakin tiny
* more control to css, less maintaining html.
initial thoughts ?
Yesterday Joe Vlcek updated aeolus-audrey-agent and aeolus-configserver
in F16 and I updated conductor, configure, aeolus-image, and aeolus-cli,
as well as two required dependencies rubygem-idn/addressable.
Michal has been keeping deltacloud up to date in Fedora, though
imagefactory/iwhd/oz could use some updates from their respective