jesus X wrote:
>
> I agree that optimizations need to be made, we all know that. But to what level
> should we drop to support wise? How slow a machine do we really need to code
> for? If we want to code for slow Pentiums with 32MB of RAM, why not 486's with
> 16? I just think it's time to cut loose the stragglers, as is always done after
> a period of time. Would you try to run office 2000 with 32MB of RAM?
My personal viewpoint: We should endeavor to be *able* to run on
anything with >=16Mb. If this cannot be achieved, 32 would be acceptable
(note that even today and even in the US, I still see newspaper ads for
computers with 32Mb).
Processor speed, as I pointed out in another post, is essentially
irrelevant - degradation is linear at worst (and non-existant at best,
because we are rarely processor-bound). Users of slow processors know
that they should expect performance to be linearly worse than if they
were using a faster processor, and find this acceptable (or they would
have bought a faster processor).
But memory usage is something else entirely, for (at least) two reasons:
1) It affects *everyone*, in terms of their ability to multitask.
2) Performance degradation is *much* worse than linear, once the system
starts thrashing.
(By "memory usage" here I really mean "working set size", because even
older systems tend to have at least 128Mb of swap that can be used for
"dead" memory use. I'm also not suggesting that we don't make use of
more memory if available - eg, for a space-time tradeoff that is only
triggered if the required memory *can* be allocated, with a fallback to
a slow algorithm if not)
> Very understandable, which is why I mentioned that it's not needed. Sometimes a
> mere hundred dollars helps things.
Many users don't know how to upgrade their computers. To make things
worse, many manufacturers specifically discourage it (eg "it will void
your warranty!")
> > I expect this memory usage issue to be resolved, and to be resolved with
> > little fuss on the part of the bloat-merchants(!) if they want people to
> > take their software seriously.
>
> This is the very attitude I'm against. I comes across as, "I'm using and old
> machine, and I demand that you cater to me, even though I'm in a small
> minority."
As I pointed out, the idea of minimizing memory usage affects everyone
who multitasks, even on modern machines. That's not exactly a small
minority. Also, there is a movement towards using second-hand computers
in places where price is an overriding factor - eg schools. As
developers of open-source software, we should attempt (as far as
possible) to make our software accessible to *anyone* who wants to run
it, rather than impose artificial restrictions.
Additionally, there are a large number of people (myself included) who
do have plenty of resources to spare, but still like their apps to be
fast and slim - just in case, for example, I need to run a large and
complex process on my machine but still have a responsive mail program.
Stuart.
