FreeLists / openbeos / [openbeos] Re: read this article...

["Michael Phipps" <mphipps1@xxxxxxxxxxxxxxxx>]

I'm with Erik. I think that Nathan was just a *bit* unfair, in that I have 
stated, publically,
about 200 bazillion times (and now I will repeat myself) that what he professes 
our
goal to be is our *SHORT TERM* goal. We want a starting point, for the future.
I really don't think that there is a person here who is afraid of rewriting 
something.
Linus rewrote the USB stuff for Linux, what, 3 times? That is *FAR* more my 
experience with engineers - never being happy with what they wrote. I know that
I am that way. The first time I got screen_saver to blank the screen, I sat 
back, smiled
for a minute, then looked at the code and said - UGH - that has to be improved.

Interestingly enough, though, I *THINK* (and I could be wrong) that the C ABI 
for GCC
has *not* changed. Only the C++. That would mean that we could build the kernel 
with GCC3 and not break any compatability. I know that Travis is using it now, 
so
it wouldn't be a porting issue. Since the kernel is where preformance is the 
most 
important, we could get some experience and practice with GCC3 and still
keep our goal of bin compatability.

Furthermore, I think that Nathan misses some points. 
Productive is a good point. I think that most of us own it and use it. 
How about a browser? Net+ and Opera may be aging, but they are sure better than 
nothing.
There are a hundred cases like this.

I agree with Nathan's desire to improve the API and to move the OS forward.
I disagree with his timing. 


>At the risk of sound snippy, we've been over this several times and have
>elected, for pragmatic reasons, to do our best to maintain binary
>compatibility for the first release -- which rules out GCC3.x for the
>time being.  This does *not*, however, rule out moving to it at a later
>date -- or even providing a parallel set of libraries which have been
>built with GCC3.x (just a "for instance"; no flames, please).  Folks
>keep insisting on thinking we've maneuvered ourselves into a dead-end by
>striving for binary compatability at this time, and that just isn't the
>case!  I respect Nathan's perspective (as well as his ability), but
>aside from whether or not we successfully achieve binary compatibility,
>this is a total non-issue.
>
>e
>
>Alex Jeppesen wrote:
>> 
>> Read this article at http://www.beunited.org/index.php?beos_reform
>> its about re-creating beos. it made a good point about how binary 
>> compatability will make it so you cannot use gcc 3.0 This is a big big 
>> problem i belive. v3.0 produces better code on many levels and i think its 
>> worth it to move on up. Anyone else? What do the OpenBeOS ppl think?
>
>