[PLUG] OpenEmbedded follow-up questions.

[Scott Garman]

Hello,

I thought I'd answer a few questions that came up at the meeting last 
night that I wasn't able to answer on the spot:

Q: Does OpenEmbedded support any devices that don't have an MMU?

I'm not familiar with too many MMU-less architectures, and my grepping 
around the machine configurations didn't lead to much. However I don't 
see any reason why OE couldn't be used to generate images for devices 
without an MMU.

Q: Does OpenEmbedded build bootloaders too?

Yes, in fact I'm seeing many u-boot recipes for various boards in the 
repository:

http://cgit.openembedded.net/cgit.cgi/openembedded/tree/recipes/u-boot

Q: Are there any tools for running compile farms for BitBake jobs?

Take a look at icecc, a distributed compile system. There are a few 
references to it being used by folks on the OE mailing list.

Q: What makes ipkg so different/better than dpkg for embedded systems?

I think this turned out to be a lot of little things that added up. One 
of the main differences is that ipkg's metadata takes up a lot less disk 
space than dpkg. Also, dpkg by itself isn't very usable without the 
apt-get front end. I may have been playing a bit too fast and loose with 
the ipkg vs. dpkg comparisons; ipkg vs. apt-get would be more appropriate.

In other ways ipkg became smaller by removing support for some things 
that apt has, such as package signing. That feature loss in particular 
created some controversy in the community, and the OpenMoko project 
forked it not long ago to create the opkg system, which appears to be on 
its way to being adopted by OpenEmbedded.

A co-worker of mine mentioned to me today that ipkg ticked him off when 
he learned that it wasn't reporting errors via its exit status 
correctly. This made it impossible to use properly within scripts (he 
had to parse the output of the command instead, blech). Apparently opkg 
resolves this bug as well.

Scott

-- 
Scott Garman
sgarman at zenlinux dot com