Re: Ada packages
Ludovic Brenta <ludovic.brenta@insalien.org> writes:
> Goswin von Brederlow writes:
>> Ludovic Brenta writes:
>>
>>> Hi Goswin,
>>> powerpc. While most of the other packages build, I think they
>>> would require at least some run-time testing before being declared
>>> good enough for Debian.
>>
>> Can you get yourself an account on one of debians amd64 machines?
>
> Even though I am not yet a Debian Developer, I don't think it would be
> too difficult. But this testing would come on top on my already heavy
> workload. Also, I may need assistance testing X11 programs such as
> gnat-gps (I've never yet had to set up an X11 connection over SSH).
ssh -X host :)
> On the bright side, when I mentioned "basic testing", I really meant
> it. If a large program such as gnat-gps (an IDE) launches correctly,
> compiles "hello world", and does not crash, chances are _extremely_
> high that it is good enough for Debian; transitively this propagates
> to libgtkada2. One of the beauties of Ada is that is is very
> portable, so, apart from glitches introduced by the compiler, I do not
> foresee any platform-specific bugs.
>
>>> However, I am more than willing to integrate your patches so that
>>> Ada packages can be supported on more architectures, if you agree
>>> to help maintain them (I am looking for a co-maintainer for
>>> additional
>>
>> I have absolutely no clue about ada and I fear none of the others
>> are ada users. Never written a single line of code or even read
>> code. The
>
> This has to be fixed. Here goes:
>
> with Ada.Text_IO;
> procedure Parallel_Hello_World is
> task type Hello_World;
>
> task body Hello_World is
> begin
> Ada.Text_IO.Put_Line ("Hello, World!");
> end Hello_World;
>
> type Task_Index is range 2 .. 8;
>
> All_Tasks : array (Task_Index) of Hello_World;
> -- All tasks start now.
> begin
> -- Wait for all tasks to terminate.
> null;
> end Parallel_Hello_World;
>
> As you can see, Ada has built-in support for multithreading (called
> "tasking" in Ada parlance). For good measure, I've thrown in a range
> type to show how buffer overflows are something never heard of in Ada
> programs :)
I like pythons 'print "Hello World.";' better. :-P
I programs ocaml a lot which also never heard of buffer overflows or
segfaults (unless you use Obj.magic or include C code)
> GCC 3.3 and/or 3.4 lack support for tasking on certain architectures,
> most notably powerpc. To me, this is quite a major problem; but users
> may disagree and be content with a mere warning in README.Debian.
>
>> intrest is more promptet by the effort to get all debian debs
>> compiled so packages are ready for testing if anybody intrested
>> shows up. The Debian-amd64 can certainly test things on a case by
>> case basis but if we start to comaintain all of the 8000 packages
>> nothing will get done anymore.
>
> Agreed. Also, you could argue that he who wants to use one of the Ada
> packages on an amd64 is probably prepared to bear with some bugs, or
> use i386 binaries. All Ada-specific packages are intended for
> developers.
>
>>> packages and architectures, BTW). I think that having GCC 3.3 as
>>> an alternative where GNAT 3.15p is not supported is reasonable. It
>>> is
>>
>> Ok, I will try to test the packages and then send the patches to use
>> gnat-3.3 on amd64 to the BTS.
>
> Excellent.
>
> --
> Ludovic Brenta.
MfG
Goswin
Reply to: