On 2012年07月18日 21:15, Richard Purdie wrote: > On Wed, 2012-07-18 at 11:13 +0800, Kang Kai wrote: >> On 2012年07月17日 17:52, Richard Purdie wrote: >>> On Tue, 2012-07-17 at 14:58 +0800, Kang Kai wrote: >>>> [Yocto #2709] >>>> >>>> OE Core Commit c9b01af84b64edba056f959c349895e0698324e6 add elf image >>>> type. This will cause hob build image failed. >>>> Add elf to hcc supoorted image types to fix this issue. >>>> >>>> Signed-off-by: Kang Kai >>>> --- >>>> bitbake/lib/bb/ui/crumbs/hobwidget.py | 1 + >>>> 1 files changed, 1 insertions(+), 0 deletions(-) >>>> >>>> diff --git a/bitbake/lib/bb/ui/crumbs/hobwidget.py b/bitbake/lib/bb/ui/crumbs/hobwidget.py >>>> index dba278c..d10c56b 100644 >>>> --- a/bitbake/lib/bb/ui/crumbs/hobwidget.py >>>> +++ b/bitbake/lib/bb/ui/crumbs/hobwidget.py >>>> @@ -89,6 +89,7 @@ class hcc: >>>> "cpio.xz" : ["cpio.xz"], >>>> "vmdk" : ["vmdk"], >>>> "cpio.lzma" : ["cpio.lzma"], >>>> + "elf" : ["elf"], >>>> } >>>> >>>> class HobViewTable (gtk.VBox): >> Hi Richard, >> >> Thanks. >> >>> Why do we have a hardcoded list here and why do new image types cause >>> builds to fail? >> After build the image, Hob trys to get list of created images. It needs >> a map from image type to real image file's extend name, such as "live" >> --> "hddimg" and "iso". >> then search the image output dir to get the list. >>> We need to rewrite this code so this does not happen and the user at >>> least gets a sensible error message. Hardcoded lists are bad... >> For the current, only image type 'live' need this map. So I can remove >> the list and check the image type is whether 'live' or not. >> But for the future, if a new image type added and need this kind of map, >> new check for the single type is needed. I think for this scenario the >> list is better. Hi Richard, > Well, I *hate* the concept that if you add a new image type, the UI has > to be changed. > > Ideally, we need to add something to the metadata that explains this > mapping to the UI. I want to see the hardcoded list removed from the UI, > one way or another. I create *Bug 2795* to follow this issue, and 2709 is closed suggested by Jessica. Regards, Kai > > Cheers, > > Richard > >