[rtems commit] New.

[Ralf Corsepius]

On 07/27/2012 05:41 AM, Sebastian Huber wrote:
> On 07/27/2012 11:33 AM, Ralf Corsepius wrote:
>>
>> How many people around here are sufficiently with the autotools to make a
>> fundated statement? I am not aware of any.

In case you are not aware about it: I am heavily using them for more 
than 15 years.

> Yes, this is one of the problems with the autotools.

Well, what shall I say?

The autotools are one amongst many tools RTEMS is using.

As with any other tool, when working with them
- users cannot avoid making yourself at least to some extend familiar 
with them.
- maintainers cannot avoid following upstreams development and to adjust 
works to upstream development, sometimes dropping backward compatiblity.


That said, if I wasn't convinced about this step being an improvement, I 
would not have applied it.

But, I am tired of
* (newcomer) users permanently complaining about "bootstrap" and having 
issues with it.
* developers facing "sporactic problems", which occasionally show to be 
using wrong versions of the autotools.


I am glad this step already has helped identifying issues with RTEMS 
buildsystem:
1. The bspopts.h.in issue you mentioned (I can reproduce it, but it is 
unfixed yet).
2. The sequence "bootstrap -c; bootstrap" has been broken until earlier 
today.
3. More smaller issues.


Finally, this step has made it possible to switch between 
rtems-4.10-branch and master in git head, without major issues.

Also, consider that, provided you have the correct versions of the 
autotools in $PATH, nothing much has changed about the work-flow.

Building with "--enable-maintainer-mode" works as before :-)


Ralf