Re: [Sheflug] Re: TopPage info for Sheffield Linux User Group

[Al Hudson <nospam [at] york.ac.uk>]

On Sat, 20 May 2000, Will Newton wrote:

> > .. assuming all platforms give equivilent services, hence wxWindows, etc.
> 
> Roughly. You have to aim at a certain group of systems. No use trying to
> port it to your TI85.

Well, yes ;)

> > 
> > I'm afraid I don't agree with that ;))) 
> 
> Well written and properly commented #ifefs can be very useful.
> Never underate having a macro system till you use Java.

I don't even use #define's in C, I'm afraid, so I still don't agree. And
yes, I have used Java.

> Hmmm, an example. DLL loading. I forget the windows call, but you can
> write code thus.
> The APIs are so similar it's not really worth writing two different dll
> loading systems.

Most examples aren't that trivial, though. And when the #ifdef .. #end
spans multiple pages, it starts getting really silly.

> > I've still yet to hear an argument that Wine adds significatly more than,
> > say, GTK, Mesa, etc...
> 
> Well if you compare GUI and graphics APIs to an OS API you won't.

I don't care what type of API it is, it still makes no difference. People
abstract the OS API all the time, it's incredibly common. 

> > If it was that easy, people would do it. People don't. You can't just
> 
> Trust me. People do. As an example try things like Unreal Tournament.
> I was impressed by the way Epic handled their ports and it's paid of for
> them.

Games aren't really all that difficult to port, though, they're kind of a
special case. Even then, I would bet that they *do* have a porting layer -
memory management, framebuffer access, etc. Porting something moderately
complicated is never as easy as that. Especially when traversing
architectures. I can show you many ansi/posix compliant programs which
aren't cross-platform, and most of the time it's not worth writing
programs which a compatible in such a manner.


> Also the fact that most XFree drivers are very poor compared to their
> equivalent Windows drivers. This is one of the areas where Linux needs to
> catch up fast.

Maybe for 3D, yes, definitely. But that's still not the main source of
inefficiency - that comes from throwing all the graphics operations up and
down the network stack. X isn't designed for speed; Windows will always be
faster than something using Xlib. DGA is a different matter, though, but
that breaks X network transparency.

> Abstracting your GUI is usually just a case of making sure all your system
> specific code is kept fenced off in it's own directory.

Depends how complicated the interface is. The example of a web browser is
a simple case - the interface is incredibly simple, that's the beauty of
browsers. If the GUI is deeply tied to the rest of the program (where the
GUI becomes interaction rather than interface) it's not as simple.

> I'd rather not. They don't seem to get things right themselves.
> Compare the man hours in Mozilla to the man hours in Opera and you'll see
> why starting afresh is often a good idea.

Errr.... Mozilla did start afresh. I don't get that last point ;)) And you
seem to be underestimating the contribution of the Mozilla guys. They're
not just doing Moz - there's Bugzilla, for example. There's a lot of stuff
there. The main thing people forget about Mozilla is, they're not building
a web browser.

> Winelib is OK for porting legacy apps, but I worry that too many new apps
> are using it as a crutch for writing crap code.

I don't think there are that many new apps writing Win code for Linux.
Porting current Win apps yes, not writing new ones. 

Cheers,

Alex.


---------------------------------------------------------------------
Sheffield Linux User's Group - http://www.sheflug.co.uk
To unsubscribe from this list send mail to
- <sheflug-request [at] vuw.ac.nz> - with the word 
 "unsubscribe" in the body of the message. 

  GNU the choice of a complete generation.