Nobu wrote:
Berger, Daniel wrote in [ruby-talk:161437]:
> Probably. I consider Win9x a dead platform, so I don't
> worry myself with it. 
Actually, we already had buried it as a development platform,
but still I'm uncertain if it is dead also as an end-user
platform.
Dilemma: Win95 is dead. Win9x is not.
Very many 98se (and, presumably ME) DLLs are NT versions
due to the convergence between Win98+ and Win2K towards XP.
So there's some software around where the author has used
the NT API and reports have been sent that it doesn't run
on 9x (they mean 95) - but it *does* run on 98+.
I've seen a few sites which say their s/w will not run on
98, but may run on 98se.
*Very* rarely, I see (for example) "XP+ only".
Developers must have to work really hard to make a program
which only runs on XP+.
In summary, I don't often find s/w that won't run on 98se.
And usa was worried that nobody would test for Win9x, if we
divide it.
Since the majority of Windows users are "one-clickers"
and there are no "one-click" stable previews OR development
versions (yet?), incompatibilities will only show up when
it's too late.
I know usa provides stable & dev. at:
http://www.garbagecollect.jp/ruby/mswin32/en/
(linked from http://www.ruby-lang.org downloads)
···
----
The incompatibilies I find in Ruby all seem to come from
the use of the command line changes (such as that >&2
redirection thingy?) in Makefiles or in scripts using
backtick/system etc. (i.e. command.com vs. cmd.exe).
Those can often be worked around (?)
Not a huge problem, IMHO.
If you want to kill 9x, the "one-click" installer must be
advertised as "2K/XP only" and it must refuse to install.
System requirements should also be advertised on ruby-lang.
Then potential users will go off to the "P"-language
websites and find that they do run on 9x - even 95.
daz
