[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Planning: Minimum System Requirements

> >> OK, I'll be more specific -- making decisions that actually MEAN
> >> something. The only spec issue that affects how the real work is done is
> >> whether '9x is supported.
> >
> > And NTFS.
> Yes, the real crux of the issue is more about whether FAT is supported.

Right.  No.

> >> Once that's decided, minimum CPU, RAM... that's just subjective fluff.
> >
> > It's minor, so what.
> The problem is not that it's minor, it's that it limits for zero gain.

The gain is not zero.

> > It flows primarily from the previous requirement
> > and it sets a starting point.  It took 10 minutes and means that nobody
> > is going to try optimizing for a 386.
> Since the base system isn't up and running yet, we are very far from
> caring about optimisation.

"Premature optimization is the root of all evil." (Donald Knuth)  Spending
hours at this point optimizng how a package compiled would be premature.
Spending a few minutes looking ahead and making some decisions is not.

> >> The
> >> minimum spec is a version of Windows that the project supports, and a PC
> >> that will run it. At the very bottom end it will be dog slow. At what
> >> point will it start being useful? A matter of opinion, and it's
> >> pointless
> >> trying to set a fixed point. Say we settle on the NT line from v4 up --
> >> some people will be able to do useful work with a 386 and 16MB.
> >
> > But that was the point!  By deciding our target "market",
> This project does not have a target market, because unless you are setting
> out to make money you do not need one. It has a technical objective. The
> target market is whoever can and wants to use the result.

This is patently wrong.  We have a market ("defn: a geographical area of demand
for commodities or services") and we'll have an audience.  The fact that they
are not paying customers does not change the fact that there is a specific
group of people whose needs we need to address.

I, for one, would rather put my time towards building a system that other
people will want to use.  That requires some planning and not just building
my own thing.

> > we don't have
> > waste time even worrying about the older machines.
> And would you like to name one way in which the project is harder if i386
> is supported? Not a vague "worrying about the older machines", try a
> *solid* reason, like the aforementioned already discussed extension and
> filename issues.

Yeah.  Running Win2k/XP on a 386.  That's much harder.  Or run dpkg on a
386.  The packaging system alone grinds it to a halt.

It's a reason.  It's not the only reason, and it may not be the best reason,
but it is a _good_ reason and in the absense of any equally good contradicting
reason holds weight towards the decision.  The fact that it supports the
other reasons, such as the extension and filename issues does not reduce
its worthiness.

I don't know why you make it sound like I'm avoiding the filename/ext issue.
That was discussed at the same time and generally agreed upon.

> > It doesn't matter
> > if some people do useful stuff on it or not; they won't be doing it
> > using this project.
> LOL. Write some code and I might believe that. Otherwise I'd say you're
> likely to have very little influence in the matter.

I guess that remains to be seen.  But since you're arguing over nothing
and making absolutely no points towards an architecture other than the
one I proposed and everyone seems happy with, I not sure what your point

If you want to build on/for a 386, go ahead.  I myself plan to use a bit more
modern system and compile for the consensus of this group.

> > Nobody is talking about hardware specs except you.  We finished that
> > discussion a week ago after a couple days
> And I'm giving my opinion now, because wasn't checking my email at the time.

Great.  You and I have now spent more time discussing this that the entire
group did originally and you have not given one reason (let alone a
compelling one) why what was agreed upon was bad and what would be a
better choice.

Stop criticizing and add some content.

                                 ( bcwhite@precidia.com )

      Relationships go through seasons.  Winter often comes before Spring.

Reply to: