Build failures w/RSB

[Chris Johns]

On 23/03/2015 9:49 am, Doug Hawkins wrote:
> Hi Chris,
>
> Thanks for your quick reply.

Pleasure.

> My host specs were hidden in the original
> msg: Linux Mint Debian Edition (all up-to-date: basically, Debian,
> similar to Ubuntu, but without all the Ubuntu-only mods) on x86_64;
> Intel i7, quad-core, 8G RAM.

Ah yes Mint and thanks for the info about it.

> My target is a TI Stellaris Launchpad
> TM4C123GLX board (the LM4F120 bsp should be fine): that's an ARM M4 core.

Nice.

>
> Re docs page: yes, that page is the same as what I found; your new link
> is better.
>

Excellent. Please update the Mint details if you feel it can be better.

> Re RSB design: didn't mean to sound so critical, sorry.

I did not take it as such. It is a valid criticism and a weakness when 
there are problems. The zlib issue is always appearing and needs to be 
solved. Of course we would like there to be no problem and I would 
expect a release to be closer to that but this is not the case.

Conceptually it is an easy problem to solve, the reality it takes 
effort. I appreciate the frustration when all you want is to be using RTEMS.

> What you've
> done is amazing and when I first read about it, that convinced me to try
> the RSB option instead of working through the RTEMS configure/compile
> issues I originally encountered.  I do understand that it can be very
> difficult to create a package like RSB and just as difficult to assure
> all possible execution paths through it will lead to success (or
> comprehensible errors).

The current complication are the churn in the tools for things like 
newlib header files, newlib moving to git, and some gdb and rtems-tools 
related errors. I expect more problem as more people use it.

> I wish I had spare time to look through it in
> more detail, maybe I could offer patches -- just not right now.

Understood.

>
> I too would prefer more RTEMS feature/function/bug fixing instead of RSB
> features/flexibility.
>

We need both, it is mostly a matter of the order and available time to 
donate.

> Yes, I've been poring over the log and error files.  Just can't tell
> what commands follow the gdb configure success (the one immediately
> after appears to be the failing command, the one that doesn't find
> python2.7).  I didn't go over the RBS source deeply enough to resolve.

The python error is usually the need to install the python2 development 
libraries.

>
> I'm now moving into compiling RTEMS directly from git.  Thanks!
>

Nice.

Looking forward to see some TI Stellaris patches and test reports.

Chris