Re: Ada platforms and pricing, was: Re: a new language, designed for safety !

["Randy Brukardt" <randy@rrsoftware.com>]

I don't see anything obviously wrong with this plan. As you probably can 
tell, I'm no fan of ASIS (it's way too syntax based for the sorts of uses I 
would want to put it toward), but that doesn't necessarily make it a bad 
starting point, especially if you're more familar with using it in practice 
than I am. You will have to be prepared to maintain the GELA ASIS code, 
because you're inevitiably going to hit bugs in it.

                      Randy.

"Natasha Kerensikova" <lithiumcat@instinctive.eu> wrote in message 
news:slrnlqagv4.i0l.lithiumcat@nat.rebma.instinctive.eu...
> On 2014-06-20, Randy Brukardt <randy@rrsoftware.com> wrote:
>> "Natasha Kerensikova" <lithiumcat@instinctive.eu> wrote in message
>> news:slrnlq7oge.i0l.lithiumcat@nat.rebma.instinctive.eu...
>> ...
>>> The way I understand the whole situation is that ASIS provider diversity
>>> would be a good thing too (though not as much as compiler diversity),
>>> and Gela is not too far from there, so pushing Gela forward would be a
>>> good thing both globally and for my self-improvement.
>>
>> That would work, of course, but all of the hard stuff (well, almost all)
>> would be in the GELA ASIS front-end. And you'd have to do frequent
>> maintenance on the ASIS stuff if you wanted to actually compile Ada (as
>> almost every bug would be in the ASIS part). It certainly could work, but 
>> it
>> would mean having to understand the GELA ASIS code in detail (not just 
>> the
>> interface). For me, I'd rather fix my own code than someone else's.
>
> Funnily, it's the first time I'm warned against re-using code instead of
> re-writing it, usually I'm warned against re-writing instead of
> re-using.
>
> The thing is, writing an ASIS provider is still frightening me, and I
> really thing it's not something I can complete by myself. And if I drop
> a re-write attempt, it's a pure loss.
>
> On the other hand, fixing Gela for Adacontrol is within human reach, and
> it's already a gain by itself, and might be the best way to assess how
> comfortable I would feel about maintaining Gela in Adacontrol and/or
> compiler (or other ASIS users) versus starting a new ASIS provider from
> scratch.
>
>>> If that succeeds, it might be a good start for an independent Ada
>>> compiler, or I might have acquired the certainty that it's not. Then the
>>> ASIS-to-LLVM-intermediate-form vs brand-new-parser-to-LLVM-IF situation
>>> will be much easier to assess, and in the meantime I will have acquired
>>> knowledge and skills that are valuable for both paths.
>>>
>>> Sounds good?
>>
>> Sounds possible. "Good" only applies if it works out and something useful
>> comes out of it. :-) Else it's just effort that could have been used on
>> something working.
>
> I meant "does such a plan sound good?". Of course you can't assess
> result goodness without knowing how everything turns out. Unfortunately,
> decisions have to be taken without knowing how everything turns out.
> However, as far as I plans go, it seems I have enough checkpoints and
> fallbacks to minimize effort loss while still moving towards meaningful
> goals, unless I'm missing something (and my question is really "with
> your experience in such endeavors, do you see something I'm missing?").
>
>
> Thanks for your insights,
> Natasha