bootstrap is slow and issues many warnings

[Chris Johns]

R. Diez wrote:
>
>>>   The process remains rather slow in any case. Is there any way to speed it up?
>
>> Normally it is just something done once or after an update. There is a
>> tool called sb-bootstrap in the RSB [1] that runs the reconfigures in
>> parallel on as many cores as you have.
>> [...]
>> http://git.rtems.org/rtems-source-builder/tree/source-builder?id=6d32fe7eb0388698ab07726229e769cd9bb30134
>
> I took a quick look at the script here:
>
>    https://github.com/RTEMS/rtems-source-builder/blob/master/source-builder/sb/bootstrap.py
>
> Unfortunately, I haven't learnt Python yet. However, it looks like it's calling 'autoreconf', so it does not use the bundled 'bootstrap' script. Is that right?

Correct it does not call the bootstrap script. Bootstrapping only 
creates the configuration scripts from the various input files a user 
creates. It does no configuration and does not perform any testing function.

> Should a tool that tests all build configurations not rely on the standard scripts that every user should run?

This is about speeding up bootstrapping and nothing more. You can select 
which ever one you wish.

> By the way, how does it run faster than 'bootstrap', which also seems to call 'autoreconf'? Or do you mean all different configurations are built in parallel?

The bootstrap script runs a single autoreconf at a time in a serial 
manner. This command uses pyhton's threads to run one autoreconf per 
core on a multicore machine. On average it should divide the time by the 
number of cores.

> I was looking at speeding up just the configuration I am using at the moment. If that method is faster, maybe it could replace the standard 'bootstrap' script.

The intention is not to replace the bootstrap script in the RTEMS 
source. It provides a simple self contained way to bootstrap which is 
faster. This command needs a number of modules present in the RSB so I 
do not think it is practical to include it in RTEMS.

Chris