Re: GNAT 3.15p transition plan
Brian May <email@example.com> writes:
> On Sun, Mar 02, 2003 at 07:06:53PM +0100, Florian Weimer wrote:
>> I'm going to recompile the existing Ada packages in Debian using GNAT
>> 3.15p, at least on x86.
> Why not aim for compiling all Ada packages with gnat-3.2 instead?
GNAT 5.0 (GNAT in GCC 3.2) is not maintained in the FSF tree, and bugs
are not fixed. It has quite a few known regressions with respect to
GNAT 3.14p/3.15p. There is no ASIS, no FLORIST (okay, no regression
for Debian), and so on.
> Or is that the same thing? (I notice that gnat-3.2 package depends on
This is a misnomer. I don't think the "official" libgnat3.15a is
binary-compatible with that library (if it existed; the DSO is
currently missing, AFAIK).
> gnat-3.2 did have some bugs that I reported upstream, eg. sin/cos not
> working. IIRC I was told that these bugs were fixed in the CVS for
> gcc-3.3. I haven't tested it myself.
There are more bugs which haven't been fixed. The last substantial
code changes in the FSF tree occured about a year ago. Internal
development at ACT does happen, but we haven't got access to it.