RFC: Drop toolchain support for rtems4.8, rtems4.9 and CentOS5?

[Chris Johns]

Oyake, Amalaye (3496) wrote:
>
> with respect to older toolchains, I would check if older Space missions
> are using the older builds. I think that ESA is baselined rtems 4.8 or
> some variant thereof ... It would be difficult to do some software
> maintenance/update on an older spacecraft if the toolchains needed are
> gone.

I see 3 separate parts to this and I wish to clarify them. The first is 
the configuration of the tools for a specific architecture and release 
plus the expected test results. It is the role of the RTEMS Project to 
define this and to provide a public repository of this information. Git 
is currently been used to hold this information and we are starting to 
fill in the details. It is a work in progress. The second part is the 
host platforms supported and packaging. What happens here are volunteer 
efforts, sponsored efforts or private efforts. The ability for the RTEMS 
Project to cover all possible cases is impossible and we do not wish to 
limit what our users want to use. In the case of RPMs, Ralf volunteers 
his efforts and so he is able to decide to manage that as best we can. 
An organization like ESA is able to select a path and manage it, or they 
can contact Joel at OAR and arrange support, or they can use the results 
of the volunteer efforts. Each as pros and cons and it is up to each 
user or organization to select a path that suites them. I do not feel it 
is up to me to dictate a path nor do I see the RTEMS Project doing the 
same. The third part is testing and test results. This is something we 
are starting to look at integrating into the data set that defines a set 
of tools. It is not clear how this will happen and what it will look 
like. The main point is to provide a way for a user or organization 
determine if the tools they are using meet the define standard. This 
last part is complex and currently a goal we have.

>
> The thing to note is that Legacy Hardware pretty much does one function
> well and stays in use for a long long time ...
>

Yes this is understood and we are mindful of the special demands this 
places on the project. It is understood the host operating systems and 
hardware we use moves much much faster than some of our user's projects. 
In my view this is a good thing how-ever obsolescence is an important by 
product that needs to be managed [1]. We are attempting to address the 
base needs and to manage them better first. The difficult part is 
meeting everyone's needs and the time needed to see how we are going.

Chris

[1] I once had to modify an archived RTEMS project and decided to use 
the carefully archived Linux distro disks how-ever the Linux kernel only 
knew about VL-bus, ISA bus and something else I have forgotten and the 
only hardware I could find was PCI bus. I used a new Linux distro, ran 
the single shell script to build the tools and a couple of minor patches 
downloaded from the net later I had verified the tools and made the 
requested change.