A timely tale...

Yesterday morning got an email from Rob that read:
"I tried to phone but either you were spending some family time, or crashed".
It was neither. I was running about the roads begging juggernauts to run me over.
Why? I had just deleted 4 days worth of work and then reformatted the drive. True!
Unbelievable I hear you say, but no, I did it. And what was worse it was intricate
work, making the final tweaks to Fant 5.10. Anyway, luckily I must have a photographic
memory as I knew almost every line of every bug fix and rebuilding what I had done 
took about eight hours. What made it worse was I backed up the whole wrong disk to
CD before the reformat - so I didn't make just one mistake, but two on the trot!

Just a sobering little story in this time of switching to HFS+ and possible mass 
formatting. Back up and CHECK you have backed up the RIGHT drive before reformatting 
(Yeah, I know "TOSSER!").

Testing, testing. Fant510.
OK, you'll like this. Hopefully, by the time you read this, in the "For the Brave" 
area you'll find lots of new stuff. Anvil 2, Fant  510 and LXT. The good news is
all this stuff is fat, so you'd expect it to be faster than V5.00. Well it is, with 
Fant 510 being especially faster than V5 - up to, and get this, 8 times the speed of 
Fant 5 on big projects. It's all been done with LXT (no, we don't use anybody elses 
stuff) and I'd bet anybody that a half decent coder could beat any native C compiler 
in about the same time with regards to the speed of the native code using LXT. For Fant 
5 to rebuild the native version of iteslf  (through LXT) was taking over 1500 seconds 
(Apple's EMU). Fant 5.1 does is it 223 seconds. 

Of course the 68K version rebuilds in about 30 and even on small projects like our 
demo, we rebuild the native code in five times the speed over Fant 5.

Anyway, you'll find Anvil faster too - it chroma codes at four times the speed of V1,
or twice as fast if you were using speedemulator with Anvil 1.

We've released this stuff now, rather than when all the new features are in simply
because it will improve your productivity. In return we ask you report bugs and 
feature requests.

LXT is something you should definitely take a look at. Even if you hate 68K code,
you'll find the OS interface macros make your life much easier. The same interface for
both 68K and PPC - plonk your parameters in registers and call the macro. Because
these macro's use a modified Xcall you can also push and pop onto the stack with 
impunity. These macros (same 250 of them so far) are tried and tested, they remove
the possibility of error in your code. If you miss or forget a parameter, you'll be 
told. If you find a macro for a call you want is not included, just add it - the macros
are simple and self explanatory as to what you need to do. We would also ask you send 
a copy to us so we can maintain and update the collection.

If you need to convert 68k to PPC, you will find LXT makes your life a lot simpler.
It directly handles translating the code, data and data space to a PPC executable.
You will probably need to make some minor source changes, but it comes with a pretty
useful manual (LSA0500), and of course you can always drop us an email and we'll sort you 
out. LXT NEEDS Fant 5.10 and won't work with V5.00. LXT is also a pretty nice language to
write in from scratch if you want a lot of control over the machine and a fast fat app.
According to how well Rhapsody is accepted, we do envisage auto translation of LXT projects
to Rhapsody. A86 is also a possibility, but it's not a high priority on our hit list.

If you deperately need to convert 68K to A86, pay us and we'll produce an LXT translator
and OS I/F to do it. Why hand translate and introduce all those bugs?

(BTW, if the software isn't there yet its because:
Currently most of the phone lines around the Towers are covered in icicles. Every time
one falls off, it makes this pinging sound that crashes all the Wintel boxes down
at the ISP!)


StuChat
My, but it's been a long time :-) But I'm back! Bigger,bolder and upgraded. Yes,
it's Stu 8.1!

A quick word about Fantasm 5.1 - it's a work in progress, this is why it creates a new prefs folder (lightsoft
 test), just in case it trashes Fant 5's prefs, so treat with caution. We've done internal
testing, and a few other people have had it for a few days, so we believe it to be 
reasonably solid. You'll need to enter your license number into Anvil2, and 9999999 
won't work.

Now here's a sad tale for you. Last night I was speaking to Rob on the phone. I said it 
would be cool if a project could have some persistant variables associated with it.
Rob, as quick as a flash of lightening, immediately said "K&R page 233 - Predefined names".
Is that sad or what? I mean, I live coding right? But that response was just too 
fast. It's frightening. There wasn't a second delay (and it's a true story folks, I'm not 
making this up). Poor Claire. (K&R is the C book, it's really called "The C programming
Language" but nobody calls it that. It's called "K&R" because the people that wrote it (the book and 
the language) are called Kernighan and Ritchie. Or Bry and Den to their mates, but "B&D" just wouldn't 
work). 

The number of the wot?
Would you release software with a build number of 666? Thought not. We were incredibly
tempted to get a witty comment out of Anvil 2 if the build number clicked over to this,
but then thought....

BSS
Picked this up from the comp.compiler newsgroup - it's something thats been bothering
me for a long time:

[I've heard a variety of explanations but here's the one I belive:
Back in the 1950s, the IBM 704 and its descendants up through the 7094
had an odd addressing scheme.  Although most arithmetic was
sign-magnitude, the index registers were two's complement and the
contents of the index register was subtracted rather than added to the
base address in an instruction to get the effective address.  Hence a
lot of programs, including everything written in Fortran, stored their
arrays backward.  The FAP assembler had two pseudo-ops to reserve
storage, BSS for Block Started by Symbol and BES for Block Ended by
Symbol.  The backward addressing has long since disappeared but BSS
lives on.  This explanation is credited to Ritchie in the Unix FAQ
that's floating around the net.  Alternative explanations welcome, but
if they don't predate 1960 and have documentation, they're not
persuasive. -John]

Also big thanks to Kevin MacDonald who sent me this:
[as for the meaning of 'bss', it has meaning to the PDP/11 assembler,
which was the machine that ushered Unix into the mainstream in the
early 1970s. 'text', 'data' and 'bss' have become popular terms
because unix uses them, unix uses 'C' and many programmers--on many
non-unix platforms--use 'C'.]

Apple
What would a StuChat be with some reference to the Mothership :-)

So (this is the weekend just gone), there we were sat in this nice little pub at the 
top (literally!) of Lincoln. Rob (and Claire) were try to educate moi about "Art". Now 
my opinion of art is that I really can't see the point of it - I mean, where's the 
creativity? More often than not, what you see is just a copy - best left to a Xerox 
or scanner IMNSHO. Rob and Claire don't see it that way.

Anyway, there we were drinking 
a nice cuppa in this pub. Rob was going on about Pre Roulette or something
similar, and I was thinking "How the hell did Apple's market share jump 47% in Europe
last year?" According to some figures I read on t'NET sometime last week (might've been
by IDC or some similar "produce nothing but boy are we important company", bit like 
Microsoft really<G> (Can I say that?, Yeah 'course I can. Chill out Bill.)) Apple's
market share did just that. Now that's one big jump. Then I looked at our orders for 
the last month, and there are an awful lot of European ones.  This is good from our 
point of view - cheaper postage and no silly green stickers to fill out. You can get SO
sick of filling those out - I mean, do they really expect you to be honest so the
poor customer can get done for a load of sales tax? Nah, course not. Anyway, how did
Apple's share jump so much? Has Europe suddenly become pro Apple? Dunno. Maybe a lot
of it has to do with these new 603 boxes (oops, sorry, I mean GEE THREE). Granted
they are fast, but it's not down to the processor so much as the decent bus speed
and cache. Anyway, maybe the [market share] figures have been "diddled"? Maybe it's 
true. Dunno.

[BIG FANFARE]
Here's today's tip for Steve:
Get 3D hardware on the motherboard as standard. License the Glide to Rave convertor
from whoever it is (IXMicro?) or write your own (faster!) one and make that a standard 
part of the OS. Then bung a load of game developers some dosh (we think you may be 
doing this already - Tempest 2000?) and get them games over here! Really, until
cheapo 3d hardware is a standard "Macism" we ain't going to get anywhere near Tomb Raider
or the size of the PeeCee game market no matter how fast the processor. Oh, and give 
us decent edge matching in that will you? And don't think expensive - think cheap - 
any old 64 bit 3d chip will do, just as long as it'll do rave, glide and openGL. 
Bung it on new motherboards and also produce a cheapo PCI (and Nubus) card and flog it 
to us faithful. 

Then, finally, if you bundle Quake, and do a whole G3 Mac avec el 3D HW for a grand, you 
will be made a Saint most pronto. 

Simple.

What's that? TriMedia? Well we've been listening to that for 2 years now. Haven't seen it
though. My Mac is less than 2 years old (and it had better run Rhap right?) and I don't
see any Trimedia chip in it? Just recently I spent close to 200Ukp on a graphics card
and Trimedia or anything even remotely Dutch was not an option.

Oh, and beeteedoubleyew, "We love you Steve!!!!!" 

And no, we don't want your y-fronts, honest. No, really, that's fine.


Gotta Love it! (Or idiot of the week award)
http://www.zdnet.com/zdnn/content/pcmo/0120/272453.html


Wrap
Well there you go. Hope you found that fun and I didn't annoy anybody. If I did
you can always switch channels or just press the "off" button.

Just a short note. After this was written it was decided  that we would need to 
take off line all test executables and make them available via email only.
Why? The usual sad tale of a cracked demo license. Sorry guys'n'gals, but now we
need you to email us your licence number and we'll email the latest goodies back
to you.

Till the next time,
Code on!


Late breaking news: It's official. A brain scan (carried out by "Excellent Cat" Fluff) on Patch revealed a complete vacuum (so she says). There is nothing there. We'll move him to packaging.

Next!


Send mail to Stu

Back to Stu's Page Top Level

Back to Lightsoft Home Page