Re: Commercial ada compilers vs gcc ada compilier

[Simon Wright <simon@pushface.org>]

Shark8 <onewingedshark@gmail.com> writes:

> The legal-complications are (IMO) due to there being three 'versions' of GNAT:
> 1: AdaCore's Community edition, which has a GPLed runtime.
> 2: AdaCore's Pro edition, which does not.
> 3: FSF's edition which has runtime- and generic-exception [IIRC] to the GPL.

I don't know what the legal position is with licence terms written in
each individual file vs licence terms specified elsewhere.

I no longer have access to any Pro editions, but

* the FSF sources contain the runtime exception
* the CE sources are (aside from a few differences caused by them coming
  from different commits in AdaCore's source tree) identical to the FSF
  sources but having the runtime exception caluses edited out of the
  licence header.

from which I deduce that there's really only one source tree, with
different releases.

So I'd expect that the individual source files of the Pro runtime
contain the runtime exception. There are (were) additional guarantees
from AdaCore about customers' abilty to use the code in proprietary
systems.

Now, back when AdaCore were removing the GMGPL from their CE-equivalent
libraries, they hadn't got round to the wheeze of programattically
stripping out the GMGPL terms from source files, and Robert Dewar
claimed that a notice in the source tree to the effect that, whatever
the individual files said in their headers, the library as a whole was
released under the pure GPL.

I always thought this was a dodgy argument, given that FSF's guidelines
for how to release your code under the GPL say and said that you should
put the licence terms in each source file
(e.g. https://www.gnu.org/licenses/gpl-faq.en.html#NoticeInSourceFile).