Quote
> >> >DOS is - unfortunately, as I think - almost dead
> >> >and Windows has already its own API's since a long time.
> >> Which only runs on Win9x. Currently there is a long thread in FPC lists
> >> about how bad NT supports DirectX :-)
> >I tested only few games under W2k, but ALL of them worked. That's
> >a good sign for me. BTW, some of them even worked better than under
> >W9x.
> I said NT. But I tried some older games under W2k too, but had a lot of
> problems. So probably only the newer games work under W2k?
For DOS games yes. NT is not DOS and the NTVDM isn't DOS either. NT does not
claim full DOS compatibility. So why do you expect such just if it looks the
same.
Linux and the DosEMU emulator doesn't allow direct hardware access for games
either. See the NTVDM as a DosEMU for NT which is a much more correct view.
The only DOS game working really perfectly (with SB sound) I know of is
Dynablasters/Bomberman (if someone cares).
For Win9x games: Most older games check for the OS version but not for the
availability of the DirectX API. So blaming the OS because of the program
doing illegal version checking, e.g. bad coding ? As you've read in the
mailing list the DirectX 5 patch for NT works fine (although it is
unofficial). But still DX 3 fully works there.
That's why older Win9x games don't work with Win2k (NT 5) either. For some
games there are patches available.
Quote
> >> >I could give you some hints for more success: change the
> >> >old-fashioned BGI-design of your unit, make it more flexible.
Agree here. All 'modern' OS'es are event driven and their biggest advantage
is that more than one app may be accessing the same device at another time.
The first time he let allow the user press the Alt-Tab key GraphiX will bomb
out irrevocably. Even at the moment GraphiX depends that the pointer to the
video memory does not change during the whole program - which is not a
recommened practice (actually it's on the list what-you-should-never-do).
Just because GraphiX has no way to tell the user such changes and GraphiX
does not interact with the API after initialization has been done...
The current design does not allow device sharing. On the Win32 platform you
may come over it by disabling the task switching keys. This can't be done on
Linux without root permissions (and tell me one real Linux user which allows
a 'fun program' having root access ?) and even there it's not considered
good practice. Apart from that why do I have an OS being capable of running
several programs at once ?
GraphiX is a huge monolithic block. Everything depends on everything else
which makes it quite hard to maintain (I believe so) or understand as
somebody interested in it. For example I (and several other people) tried to
extract the image loading / saving stuff which would be very interesting for
other purposes too - but after a few hours it was clear that a complete
rewrite would be easier and less tedious work. Too less comments, too less
documentation and changing the slightest nuance of the code breaks it at
twenty other places.
Quote
> >> Afaik Graphix uses LFB which is probably faster tahn directX.
> >I meant the API. He still uses the simple BGI-style. That's
> >combined with alot of disadvantages. I saw how he copied alot
> >of the functions of GrafX2. Yes, he's got now similar routines - but
> >they aren't as flexible as they could be.
> >BTW, you can also have direct screen access under DirectX - so
> >it's almost as easy to use as the LFB and as fast.
I disagree. DirectX drivers usually serve as extended graphics BIOS. That
means they enable one or two hardware register contents (e.g. PII Write
Combining) which DOS does not. This quite pleasant feature also causes
Linux' drivers not to detect devices (due to very paranoid behaviour btw)
when it is booted via e.g. loadlin after a Windows was running.
But it can be done in another way too - for example BeOS starts perfectly
right out of the box.
Well, in effect the difference between running GraphiX under Win9x DOS box
and pure DOS is 3 times as fast LFB write access on a PPro system.
Quote
> You can see that on Win32. Quite a lot of FPC Win32 users also use Delphi
> (and bought it). They simply use FPC on the side for things that
> Delphi can't do, and invest some extra time to keep their code compilable
> with both Delphi and FPC. (e.g. inherit from GUI widgets on relatively
late,
> not using dynamic arrays etc). In return their codebase is more flexible
and
> portable.
I guess that far the majority of FPC users are still doing GO32V2
programming. Either for compatibility reasons (the software _must_ run as, I
think, you said) or for being the least complex in the beginning. The rest
is probably equally divided by Linux and Win32 - but only for Console Apps,
e.g. stuff originally written for DOS but ported to Win32 because of LFN
etc. Small tools which make life easier.
The graphical side of both OSes is in a experimental stage on Linux but
Win32 GUI programming is a very rare exception, may it be DX / GDI
programming.
Quote
> This also selects the users FPC gets: it gets users from Delphi that do
> things for which Delphi isn't flexible enough, IOW the top 10%, not the
Rather which don't fire up a GUI just for fourty lines of code. Delphi can
basically do the same things as FPC - and a Delphi programmers doing a small
console tool with FPC doesn't make him a FPC programmer. And as you have
outlined they keep their sources compilable at both platforms so they won't
use FPC for its' extensions. Just because the commandline compiler has a
smaller turnaround time.
Quote
> people that drag and drop make the zillionth screepsaver.
> This is quite important.
You won't need to install them all ;)
Quote
> >Some comments to the other postings: we are speaking as a minority.
> >I don't know much companies which use these "free" libraries we
> >produced.
> Who said something about companies? Why would we care for companies?
Which company wants to give away its sources (technology) for free ? Still
they need to earn a living for that...
If they did most probably nothing would change, at the very least in short
to middle term. Why would I download the NS sources ? It doesn't change for
me whether it's available or not. I will certainly never have a look at it
for any reason and even change something (or anyone else). NS keeps
implementing propretiary extensions and nobody simply makes an alternative
free browser out of them. In effect Navigator development kept at Netscape
and still it crashes at simple style sheet tags.
It makes no difference. The browser did not get better just because it got
'available for free'. As a user I don't have much gain if I can say I
*could* change something here and there. It's simply not the way it goes.
(Impressing people with that 'argument' is another thing)
Regards,
Thomas