[OE-core] Multiple MACHINE building is broken in OE-core?

[Mike Looijmans]

On 02/01/2013 11:29 AM, Burton, Ross wrote:
> On 1 February 2013 07:51, Mike Looijmans <mike.looijmans at topic.nl> wrote:
>>> A concrete example might help.
>>
>> Would be glad to comply, if you care to explain what you need. What should I
>> do, which logging to collect (and maybe how)?
>
> To start with just the two machines you are changing between and the
> packages/tasks that are re-ran.

How can I get a list of packages/tasks? They just flash by on screen, 
but I bet there's a convenient flag that I can pass to bitbake to get 
what I want.

One case was between "zynq-zc702" and "zedboard" using my BSP overlay on 
https://github.com/milosoftware/meta-zynq
Both boards have the same SOC, just different board configurations.

Another is when switching from settop box "et9x00" to "vusolo2" in the 
upcoming OpenPLi at 
http://openpli.git.sourceforge.net/git/gitweb.cgi?p=openpli/openpli-oe-core;a=shortlog;h=refs/heads/upstream
The latter failed so miserably that I had to remove the whole "tmp" 
directory and start over to get it to build at all for either of them.


 > As Martin said, switching machine
> will mean a new sysroot so you'll see lots of setscene tasks running
> for every package to start with, but that will only take a minute.

Would using INHERIT += "rm_work" in my config files be problematic here? 
If I need to preserve all the work items, I'll go and buy a bigger 
harddisk, but it's more efficient to just clean up.

Mike.

(sorry, forgot to click the "reply to list" button in my previous response)